Beating The Grünfeld PDF

Beating
the Griinfeld
Beating the Griinfeld
Beating the Griinfeld

ANATOLY KARPOV
Translated by John Sugden
B.T. Batsford Ltd, London
First published 1992

Anatoly Karpov 1992
ISBN 0 7134 6468 2
British Library Cataloguing-in-Publication Data.
A catalogue record for this book is available from the British Library
All rights reserved. No part of this book may be reproduced, by any
means, without prior permission of the publisher
Typeset by Lasertext Ltd, Manchester
and printed in Great Britain by
Dotesios Ltd, Trowbridge, Wilts
for the publishers,
B. T. Batsford Ltd,
4 Fitzhardinge Street,
London WIH OAH
A BATSFORD CHESS BOOK

Adviser: R. D. Keene GM, OBE
Technical Editor: Andrew Kinsman
Contents
Preface
Seville Variation
11
Exchange Variation with 7 c4 - other systems
41
Modern Exchange Variation
52
4 Russian System
108
4 f4 System
135
Fianchetto System
1 52
Index of Variations
1 87
Preface
I have to admit that the title of this book has an air of sales-talk
about it, as similar titles generally do. Anyone trying to think up
a recipe for beating this or that opening will scarcely succeed, and
the Griinfeld Defence is no exception. But seriously: to chessplayers
whose repertoire includes the Griinfeld, the author imagines that
the present work will be of considerable interest and use. Hence
it is indeed quite possible that the book will help many readers
to score wins - with Black if they are adherents of the Griinfeld,
and with White if they are looking for a way to combat this
defence.
The book consists of a collection of games (or fragments of
games) which reflect the contemporary state of Griinfeld theory.
It is constructed around twenty-five paradigms of play by con
noisseurs of this opening (like many authors I have a weakness
for round figures), but the total number of examples is about ten
times higher! The notes to each of the twenty-five principal
games constitute a thorough discussion of the currently popular
variations. As a result, all the most fashionable systems occurring
in grandmaster practice in the last few years have found their way
into the book. The selected games are not arranged chronologically
but grouped according to themes. For this reason, the freshest
examples - those from the recently concluded duel for the world
chess crown, which incidentally are the most fully annotated - do
not form the culmination of the book, but are placed in the middle
of it . . . .
This book is written in the same format as the four-volume
work probably already familiar to the reader : The Open Game
(Semi-Open Game I Closed Openings I Semi-Closed Openings) in
Action. That is to say, the scores of all the principal games are
given in full, and in analysing the opening the reader will mostly
be able to study its relation to the middlegame or even the
8 Preface
endgame. In some games, the later stages are of independent

interest and consequently receive fairly detailed notes. The informal
layout of material has permitted the author to select contemporary
games according to his own discretion (in contrast, say, to an
opening monograph, where you also have to include systems that
have not been used in practice for a long time).
Whereas the four-volume work just mentioned covers the
development of theory in the period 1984-7 (the last-but-one
World Championship cycle), all the principal Griinfeld games in
the present book are from the later period 1988-90, including the
most recent World Championship match.
In my encounters with Gary Kasparov, the Griinfeld Defence
has figured prominently. In our last three matches the most varied
systems were tried out, and in our preparations we utilised all the
most important theoretical material. As a result of these matches,
the theory was in turn substantially enriched. Many ideas employed
for the first time by Kasparov and myself have seen an onrush of
further developments. Suitable examples of this are included in
the book, while practical sources prior to 1 988 are incorporated
in the notes to the principal games.
I should mention that I usually play the white side of the
Griinfeld while Kasparov plays Black, and that both of us strive
for victory (not only when playing each other). Such is the nature
of this sharp and uncompromising opening, in which playing for
the draw is inimical to both sides - although, to be sure, our fierce
contests have often ended peacefully. For those who like statistics,
let me give my overall score against the Griinfeld in each of
these matches. London/Leningrad 1986 : +3 = 6. Seville 1 987:
+ 2 -1 =7. New York/Lyon 1990 : +1 = 3. As you can see, we
have played almost an entire World Championship match of
Griinfelds - with a definite plus score in my favour. If you like,
you may conclude from this that I have unearthed the secret of
Beating the Grunfeld . . . .
Obviously it is not only the World Champion and ex-champion
who repeatedly contest this popular opening. It belongs to the
repertoire of many prominent players. Suffice it to mention such
names as Vassily Ivanchuk, Boris Gelfand and Jan Timman. Some
of the valuable discoveries by these super-grandmasters will be
found in this book.
In the course of working on the text, I have utilised a great
Preface 9
many sources (Informator, books and magazines), and this is
reflected in the sheer quantity of references to games by masters
and grandmasters. In the majority of cases where a variation or
individual move is of major significance, its originator is mentioned.
In conclusion, I must thank Soviet Master Evgeny Gik, my co
author in many books, for his help in preparing the manuscript.
Anatoly Karpov
1 Seville Variation
Game No. 1
Kuzmin-Henkin
Moscow 1989
I
2
3
4
5
6
7
8
9
10
d4
c4
.!Llc3
cd
e4
be
c4
.!Llel
e3
0-0
.!Llf6
g6
d5
.!Llxd5
.!Llxc3
g7
c5
.!Llc6
0-0
g4
The game has followed the main system of the Griinfeld. I

would just remind the reader that in the event of 1 0 . . . cd 1 1 cd
.!Lla5 1 2 d3 e6 1 3 d5, theory states that White obtains a
dangerous attack for the exchange.
ll
1
w
f3
.!Lla5 (1)
12 Seville Variation
But now Black can answer 1 2 .td3 with 1 2
cd 1 3 cd .te6,
and the exchange sacrifice (14 dS) is less dangerous to him, since
in several lines he has an important queen check on b6 (the details
can be found in any reference work on the Griinfeld). However,
in this situation White's exchange sacrifice is by no means forced.
We shall later acquaint ourselves with a recent game with 14 .l:l. c l
instead (Game No. 6 , Yusupov-Kasparov).
. . .
12
.txf7+
So International Master Kuzmin, one of my seconds, has

decided to participate in the theoretical debate launched by
Kasparov and myself in Seville. After that 1 987 match, Kasparov
stated that the plan chosen by White with 12 .txf7 + was
unpromising. But to judge from the way the play went in the five
games in question, it cannot be said that Black easily solved his
opening problems. And in the post-match duel to which we shall
presently turn (Game No. 2), he was thoroughly routed.
The pawn structure that now arises gives White every reason
to count on the initiative, besides which he has an extra pawn. But
then again, the position is highly dynamic and may very well suit
the taste of the player of the black pieces.
Anyway, interest in this variation has not died down in three
years, and numerous games enriching its theory have been played.
Most of them will be mentioned in this book, over the first five
main games. Of course, to some extent it will also be necessary
to refer back to the Seville games.
12
13
14
fg
..txfl
.l:l. xf7
.l:l.xfl +
Let me emphasise that the main feature of the position is not

the extra pawn; the freedom of Black's game compensates for this
minor deficit. White's basic plan is to block up the enemy bishop
on g7 by means of the pawn chain c3/d4/e5/g5/h4. Black will rely
on tactical devices to enable his bishop to escape onto the h6-c l
diagonal.
14
"it'd6 (2)
This queen sortie occurred in three games in the Seville match
and two of our later encounters. It has also been played in many
other games in recent years. Let us nonetheless mention some
alternatives.
In the ninth game in Seville, Kasparov preferred 14 ... cd 1 5 cd
,
Seville Variation 13
2
w
11V b6, and after 16 g1 11Ve6 17 11Vd3! I returned the pawn while
keeping all my positional trumps. The offshoots arising from 1 4
. . . c d 1 S c d will be examined i n d e tail i n Game No. S (apart from
1 S . . . "it"b6, the moves 1 S . . . "it"d7 and 1 S . . . eS have been seen).
The Seville match was not yet over when the new move 14
"it"d7 occurred in a game Chernin-Gavrikov, Lvov 1 987. That
game proceeded : 15 de J: f8 + ( 1 S . .. "it"xg4 16 f4!) 16 g 1 "it" xg4
1 7 f4 ( 1 7 "it"d3 c6 1 8 h3 is not bad either) 1 7 . . . "it" xd 1 + (but
not 1 7 . . . J: xf4? 1 8 .txf4 "it" xf4, on account of 19 "it"d8 +) 1 8 : xd 1
.txc3 1 9 d S (White gains nothing from 1 9 e6 : c8 2 0 J: d7
f7 2 1 gS+ e8 22 J: d3 .tb4, or in this line 2 1 .tgS? .tf6!
22 .txf6 xe6) 1 9 . . . .tf6 20 .th6 (the advantage is with White,
but Black manages to hold on) 20 . . . J: e8 (but not 20 . . . : f7 2 1
xf6 + : xf6 22 e S :rs 2 3 g4, and White wins) 2 1 xf6 + ef 22
J: d7 : xe4 23 J: g7 + h8 24 : c7 g8 2S J: g7+ (2S .td2 c6
26 : xb7 J: e7) 2S . . . h8 26 : c7 g8 27 J: g7 + h8 28 l: c7,
draw. Gutman suggests 20 g4! : e8 21 J: b 1 a6 22 f2, followed
by e2 and .td2, with the initiative.
In answer to 14 . . . "it"d7, a more logical move seems to be 15
g5 (3), as played in Karpov-Gavrikov, European Speed Chess
Championship, Spain 1 988.
Despite the 'non-serious' nature of the contest, the game is of
considerable interest : 15
"it"e6 16 eS! "it"c4 ! 17 g1 J: d8 18 "it" e 1
c6 1 9 .tf2 a 6 (wouldn't 1 9 . . . b S ! ? have been better?) 2 0 a4
aS 2 1 h4 .tf8 22 de! b3 23 .l:!.b l ! (more precise than 23 J: d 1
: xd 1 24 "it"xd 1 lbxcS) 2 3 . . . e6 (but now 23 . . . xeS loses t o 24
: b4 "it"dS 2S J: d4) 24 c6! be 2S d4 lbxd4 26 cd "it" xa4 27 "it"c3
(White has an obvious endgame advantage) 27 . . . J: c8 28 hS! gh
29 "it"h3 J: e8 30 "it"xhS J: e7 (30 . . 'lf c2 3 1 J: b7) 3 1 g6 11V c2 32
ooo
3
B
gh + .l:l. xh7 33 11fg4+ h8 34 .l:l. f l 11ff5 (34 . . . 11rd3 35 -*.h4!) 35

11' xf5 ef 36 -*.e3 .l:l. h5 37 g4 ! .l:l. h3 38 .l:l. xf5 .l:l. xe3 39 .l:l. xf8 + g7
40 .l:l. f4! (the rook ending is won for White) 40 . . . a5 4 1 f2 .1:1. b3
42 .l:l. f6 .l:l. d3 (42 . . . .l:l. c3 43 e2 a4 44 d2 etc.) 43 .l:l. d6 a4 44
e2 .l:l. g3 45 .l:l. xc6 .l:l. xg4 46 d3 47 .l:l. a6 f7 48 c4 .l:l. g3
49 d5 1-0.
From diagram 3, a game Gligoric-Popovic, Yugoslavia 1 988,
continued differently : 15
.l:l.d8 16 g1 e6 (an evident improve
ment is 1 6 . . . cd 1 7 cd h8 1 8 "ifd3 ll:lc6 1 9 .l:l. d 1 ll:lxd4 20 ll:lxd4
e5) 17 .1:1. b1 ll:lc4 ( 1 7 . . . cd 18 cd ll:lc6 19 d5 ed 20 ed 11fxd5 2 1 "iWxd5
.l:l. xd5 22 .l:l. xb7 is good for White) 1 8 j.f2 b5 1 9 11rd3 a6 (White
can now obtain the better game with 20 h4 ll:le5 2 1 11fh3, but
Gligoric prefers to move his other rook's pawn) 20 a4 ll:le5 2 1
11fc2 (now 2 1 11fh3 ll:lc6 2 2 ab a b would lead t o unclear play - but
not 21 . . . ba 22 de a3 23 .i.e 1 ! etc.) 21 . . . ll:lg4! 22 e5 b
(Gligoric gives the variation 22 . . . ll:lxf2! 23 xf2 b4! 24 de
.i.xe5!, and Black has no problems) 23 .i.g3 11fc6 24 "iWa2 ll:le3
25 ll:lf4 (White has a slight edge in the endgame, but it isn't enough
for victory) 25 . . . ll:ld5 26 ll:lxd5 11fxd5 27 "iWxa4 cd 28 cd "iWxd4+
29 11fxd4 .l:l. xd4 30 .l:l. b6 f7 3 1 .l:l. xa6 .l:l. d5 32 .l:l. a7 + g8 33 h4
..txe5 34 -*.xeS .l:l. xe5 35 .l:l.e7 .l:l. e4 36 g3 .l:l. e 1 + 37 f2 .l:l. e5 38
f3 .l:l. e 1 39 f4 .l:l. e2 t - t .
Perhaps a n even sounder answer t o 1 4 . . . 11r d7 i s 1 5 h3. This
occurred in Yusupov-Popovic, Belgrade 1 989. There followed 1 5
. . . ll:lc4 1 6 .i.f2 cd 1 7 cd e5 1 8 de ll:ld2 + 1 9 e 1 .txe5?
(according to Yusupov, unclear play results from 19 . . . .i.h6 20
11fc2 .l:l. c8 2 1 11fb2 ll:lxe4 22 "iWb3 + ) 20 .l:l. c 1 .l:l. d8 21 "ifc2!, and in
the endgame White made no mistake in exploiting his extra pawn.
In addition to 14 . .. cd, 1 4 . . . 11rd6 and 14 . . . 11rd7, Black has
. . .
one other option : 1 4

1Wc8. This move was tried out in Portisch
Korchnoi, Reykjavik 1 988. After 15 1Wa4?! Black obtained an
active game with 1 5 . . . cd 16 cd lbc4 17 .tf4 a6 1 8 g5 b5 19 1Wb3
e5 20 de 1Wc6. At this point, instead of 21 lt d 1 , White should have
preferred 21 1Wd3!? lbxe5 22 1Wd5+ 1Wxd5 23 ed b4, with about
equal chances (but not 23 . . . lbd3 24 lt d 1 lbxf4 25 lbxf4 ltf8 26
g3 .te5 27 d6 .txf4 28 gf : xf4+ 29 "'e2, with initiative for
White). Korchnoi answered 2 1 ltd1 with 2 1 . . . lt f8!, whereupon
Portisch played 22 ltd5, overlooking the blow 22 . . . 1Wxd5!. After
23 ed lbd2+ 24 "'el lbxb3 25 ab lt d8, Black gained a decisive
endgame advantage - though it took him fifty more moves to
achieve the win!
Evidently White should react to 14 . . . 1Wc8 in the same way as
if the queen had gone to d7, with 15 h3 or 15 g5. Seirawan
Kudrin, USA Ch. 1 989, went 14 . . . cd 1 5 cd 1Wc8 16 g5, when the
black queen utilised its possession of the c-file with 1 6 . . . 1Wc4.
However, after 17 "'g1 lt d8 18 1We1 lbc6 19 e5 e6 20 1Wf2 lbb4
21 lDf4 ltc8 22 ltfl , it became clear that the raid with the queen
had achieved nothing and the white centre was invulnerable. In
the endgame after 22 . . . lDd3 23 lbxd3 1Wxd3 24 1Wf7+ "'h8 25
.tf2, White had a material and positional plus, which he duly
converted to a win.
.
15
e5
In the 1 1 th match game in Seville, I chose 15 "'gl . For the

current state of theory on that move, see the notes to Game No.
4, where the completely new 1 5 1Wa4!? will also be discussed.
15
1W d5 (4)
15
-. e6 has also been seen. After 16 g5 lbc4 ( 1 6 . . . 1Wc4 1 7
"'g1 transposes into Karpov-Gavrikov, where play resolved itself
clearly in White's favour), White has to avoid the trap 1 7 1Wd3?
1Wf5 + ! which costs him a piece; but even with 17 .tf2 lt f8 1 8
"'g1 1Wf7 1 9 1W e 1 lba3 20 ltc l 1Wxa2, h e achieves nothing.
A game Makarov-Hodko, USSR 1 988, went 16 h3 lbc4 17 1Wd3
ltd8 1 8 1We4! ( 1 8 lbf4 1Wa6!) 18 . . . 1Wc6 19 1Wxc6 lbxe3 + 20 "'f2
lbxg4+ 21 hg be 22 lt b 1 cd 23 cd c5 24 "'e3! cd+ (24 . . . lt d5
25 "'e4 e6 26 : b5 .tf8 27 : a5 is bad for Black) 25 lbxd4 : d5
26 lbc6. At this point, in Makarov's view, Black should have
played 26 . . . lt c5; after 27 lt b8+ .tf8 28 lt c8 "'g7, White has
no more than a slight edge.
Quite a good reply to 1 5 . . . 1We6 seems to be 1 6 lDf4 1Wc4+ 17
'ife2! : f8 1 8 'ifxc4+ 'Oxc4 1 9 e2, with the better ending for
White. One other possibility, 16 g1 was tried in Hansen
Fereec, Aosta 1 989. Black restored the material balance, but after
1 6 . . . 'ifxg4 1 7 'iVd3 'ife6 1 8 -*.g5! : f8 1 9 h3 cd 20 cd -*.xeS 2 1
de 'ifxe5 22 -*.c l ! 'ifxa1 2 3 "ird5 + e6 24 'ifxa5 : c8 2 5 'iVd2 b5
(doubtless a more accurate line was 25 . . . 'iVb 1 26 h2 'ifc2 27
1We3 "irxa2 28 'Oc3 "irb3 29 .i.d2, with a minimal plus for White)
26 a3, his position was fairly difficult.
,
4
w
16
.i.f2
In this case 16 'Of4 is weak, since the queen gets to e4. A game
Lichak-Asrian, Leningrad 1 990, went 16 . . . : f8 1 7 g1 "ire4 1 8
"irf3 "irxf3 1 9 gf .i.h6 2 0 10d5 -*.xe3 + 2 1 'Oxe3 cd 22 cd 'Oc6 23
: b 1 b6 24 g2 'Oxd4, and Black had an endgame advantage.
16
:rs
So far, the play coincides with game 5 in Seville. In the 7th

match game, the black rook preferred to go to d8, forcing the
white queen off the central file. In later games, 16 . . . : d8
completely replaced the transfer of the rook to f8; we shall go into
details in the notes to the next game in this book.
17
gS (5)
A valuable novelty. The game in which this position first arose
(number 5 in Seville) continued 17 g1 .*.h6 1 8 h4 "irf7 19 -*.g3
.i.e3 + 20 h2 "irc4! 21 : b 1 (2 1 de is met by 21 . . . "irxg4, with
a good game) 21 . . . b6 22 : b2 (but here 22 de is sounder) 22 . . .
"ird5! 23 "ird3 'Oc4 24 : b 1 b5, and the sharp duel should have
ended in a draw (although a blunder by Kasparov eventually
enabled me to win). However, 24 . . . g5 would have given Black
good winning chances.
With 17 g5, White achieves his principal aim of shutting off the
opposing bishop. Admittedly his king is dangerously placed
opposite the black rook, but this is just a temporary problem.
5
B
17
Wf7
Gutman gives the variation 1 7 . . . "ife4 (or 1 7 . . . tt:Jc4 1 8 g1

"ife4 1 9 tt:Jg3 "iff4 20 "if e2 b5 2 1 tt:Je4, with the better game for
White) 1 8 tt:Jg1 ! tt:Jc4 1 9 ltlf3 ltle3 + 20 .i.xe3 'if xe3 2 1 "iib 3 +
h8 22 J:te 1 !, and Black's position is not to be envied. 22 . . .
J:txf3+ may look inviting, but after 23 gf "if xf3 + 24 g1 "ifg4+
25 f2 "ifh4+ 26 e2 "ifg4+ 27 d2 "ifxg5+ 28 c2, there is
no perpetual.
18
"ife1
h6
In reply to 18
'if f5 Henkin gives these variations :
(a) 19 h4, and now :
(a1 ) 19
"ifg4 20 g1 h6 (or 20 . . . ltlc4 2 1 tt:Jg3! "if xh4 22 tt:Je4,
with a distinct plus) 2 1 gh .i.xh6 22 ltlg3! cd 23 cd "if xh4 24 tt:Jf5!
"ifg5 25 tt:Jxh6+ "if xh6 26 "if xa5, and wins.
(a2) 19
h6 20 gh .i. xh6 21 ltlg3, with a clear advantage.
(a3) 19
"if e4 20 g1 cd 2 1 cd ltlc4 22 J:t c l ! (but not 22 g3?
.i.xe5 23 de tt:Jxe5) 22 . . . ltle3 23 ltlf4!, and if 23 . . . l:t xf4 24 .i.xe3
,J xh4 25 J:tc8 + , White has a considerable initiative.
(b) Another playable line is 19 ltlg3 "if xg5 20 tt:Je4 'iff4 21 g1
cd 22 cd tt:Jc6 23 .1: b 1 ! with the advantage.
In a game Kuzmin-Malishauskas, USSR 1 989, Black played a
new move, 1 8
ltlc4 ( 1 8 . . . ltlc6 is also possible). There followed
19 tt:Jg3 tt:Jb2 20 tt:Je4 tt:Jd3 2 1 "ife3 (2 1 "ife2 tt:Jf4 !) 2 1 . . . tt:Jxf2 22

...
lt:lxf2 (Kuzmin recommends 22 . . . "it'c4 + , without the exchange

in the centre; after 23 g1 .z:td8, Black has the initiative) 23 cd
"it'c4 + 24 g1 .z:td8 25 "it'b3 "it'xb3 26 ab llxd4 27 llxa7 J.xe5
1-t (Black would lose after 27 . . . llb4 28 lt:ld3 ll x b3 29 lla8 +
f7 30 e6+ !). A much stronger answer to 1 8 . . . lt:lc4 is 19 lt:lgl!
followed by 20 lt:lf3.
19
gh
Stronger than 19 lt:lg3 hg 20 lt:le4 cd 21 cd "it'f5.

19
20
J.xh6
g1
lt:lc4
An inadequate alternative is 20 . . . J.d2 2 1 "it'xd2 "it'xf2+ 22
h 1 g7 23 "it'd3 lt:lc6 24 lldl .
21
lt:l g3!
J.dl
Otherwise White is a clear pawn up. The bishop move to d2

would also have been a good reply to 21 ..th4.
22
"it'el
..txc3
23
llfi!
23 e6? "it'xf2 + 24 "it'xf2 llxf2 would lose for White.
"it'e6
23
If 23
cd, then 24 e6! is immediately decisive. On 23
lt:ld2?,
White has 24 .z:td 1 cd (24 . . . "it'xa2 25 lt:le4!, or 24 . . . "it'f4 25 de!)
25 llxd2 J.xd2 26 "it'xd2 d3 27 -*.xa7.

24
25
de!
lle1
..txeS
"it'dS
25 . . . "it'f7 or 25 . . . llf4 would be safer.

26
lld1
If 26 "it'g4 (26 lt:le4 ..td4!), Black has 26 . . . llxf2! when Henkin's

analysis goes : 27 "it'xg6+ f8 28 xf2 (28 "it'h6+ J.g7) 28 . . .
J.d4 + ! (28 . . . "it'd4+ 29 fl lt:le3 + 30 llxe3) 29 fl lt:le3 + 30
: xe3 (but not 30 e2 "it'xg2 + 3 1 d3 "it'c2+) 30 . . . ..txe3 3 1
"it'f5 + "it'xf5 3 2 lt:lxf5 ..txc5, and Black seizes the initiative.
26
ll xf2
After 26 . . . "it'e6 or 26 . . . "it'f7, Black could still offer resistance,

but now the game ends at once.
27
28
29
30
llxd5
lt:lxe2
c6!
: xeS!
1 -0
llxel
f7
b6
Game No. 2
Karpov-Kasparov
Belfort 1988
1 d4 ll:lf6 2 c4 g6 3 ll:lc3 d5 4 cd ll:lxd5 5 e4 ll:lxc3 6 be .tg7 7
.tc4 c5 8 ll:le2 ll:l c6 9 .te3 0-0 10 0-0 i.g4 1 1 f3 ll:la5 12
.txf7 + .:t xf7 13 fg .:t xfl+ 14 xfl 'it'd6 15 eS 'it'dS
16
i.f2
.:t d 8
Mter the 7th game in Seville, in which Kasparov played this
move for the first time, the rook manoeuvre to d8 became standard
practice in this position. On d1 the queen was quite conveniently
placed, but now the threat of 1 7 . f .txeS compels it to leave its
post. White gains nothing from 17 ltlf4 'ifc4+ ( 1 7 . . . 'iff7 1 8 ll:lh3!)
18 11Fd3 'ifxd3+ 19 ll:lxd3 cd 20 cd ll:lc6 21 ltlcS ll:lxd4 22 ll:lxb7
.:tdS 23 .:t e l : xeS 24 : xeS .txeS 2S ll:ld8 a6, with equality.
There are three ways for the queen to vacate the d-file : along
the d l -a4 diagonal (to a4 or c2), or with 17 'ifel. We examine 17
'ifa4 in the present game, and 17 'ifcl in Game No. 3. But first,
let us recall how the Seville game proceeded : 17 'ife1 'ife4 1 8 gS
'iffS! 19 h4 ltlc4 (he should have put pressure on the centre with
19 . . . ll:lc6 20 gl 'ife4, leading to sharp play) 20 g1 'ifg4 (20
. . . bS was more precise, as after White's next move, the black
knight will feel uncomfortable) 21 a4 h6 22 : a2! hg 23 'it'bl !
(White thus succeeds in exploiting the weakness of the a2-g8
diagonal) 23 . . . gh 24 'it'b3 'ife6 2S ll:lf4 'iff7 26 ll:lxg6 'ifxg6 (I
shall refrain from further comment until we reach one particular
critical position; this book is not the place for exhaustive notes
on the Seville games, they are no doubt well known to the reader
already) 27 'ifxc4+ h8 28 .:t b2! cd 29 cd 'ifg4 30 'iff7! .J:I. xd4
3 1 .txd4 'ifxd4+ 32 .:t f2 'ifxe5 33 : fs 'ife l + 34 :n 'ifeS 3S
hl? (although we are quite a long way out of the opening, it is
appropriate to take stock of the situation. With 3S 'iff4!, White
should come out on top : 35 . . . 'ifxf4 36 .:txf4 i.f6 37 : c4 g7
38 .:t c7 b6 39 fl . with a won ending) 3S . . . b6 36 'iff4 'ifhS (a
withdrawal that was impossible a move earlier. If now 37 'it'b8 +
h7 38 'ifxa7, the white queen is far from the scene of action,
and after 38 . . . 'ife2 Black has enough initiative to draw. But
White has another way to exploit the awkward position of the
black queen on the rook's file) 37 'iffS? (at this point, 37 .:t f3! i.f6
38 1Wb8 + g7 39 'ifxa7 1Wc5 40 1Wa6 was decisive) 37 . . . 1We2 38
.:t e l ? (here was White's third successive opportunity to win this
.
game. He should have checked first - 38 "it'c8 + h7 - and only
then played 39 :t e l , threatening 'iltc2 + . If 39 . . . 'iit h 5, then 40
'iltc2+ h8 41 'iitd l ! etc.). The game lasted another 40 moves,
but all my efforts to overcome my opponent were in vain. Although
Black was on the brink of defeat more than once in this game, we
have seen that the opening was not to blame. After 1 7 'ilt e l , Black
had the means to obtain a perfectly reasonable position.
17
'ilta4
(6)
6
B
Now after the reply 1 7 . . . b6, my queen would withdraw to c2.

You might ask what point there is in provoking a useful move of
Black's b-pawn instead of placing the queen on c2 at once. The
explanation is that in some variations (after Black's . . . "it'c4, for
example) White can play 'ilte4, gaining a valuable tempo by
attacking the knight which would by now have returned to c6.
So 1 7 'ilta4 has its plus side. But I will not offer a categorical
verdict as to where the queen feels more comfortable, although I
have tried both possibilities in my games.
17
b6
In the event of 17 .. . lLlc6 1 8 "it'b3 c4 19 'iltxb7, White has the

better game.
After 17
lLlc4 18 lLlf4 'iltf7 19 g3 lLld2 + 20 g2 lLle4 2 1 'iltc2
lLlxf2 22 'iltxf2 White would again have a plus, but at move 20
Black has the powerful 20 . . . g5 !, pointed out by Gutman.
He suggested that White should play 18 g5! himself, with the
continuation 18 . . . l:t f8 19 g1 'iltf7 20 ..tg3 lLle3 21 ..tf4 lLld5
22 l:t fl . Black's initiative has been neutralised and his bishop
imprisoned, which is to say that the opening contest is settled.
.. .
After 17
l:f8 1 8 g1 'llf7 1 9 .*.h4! ll:lc4 20 'ilb3, the bishop
is again unable to break out and free itself: 20 . . . .*.h6 21 g5!
.i.xg5 22 .*.xg5 'ilf2 + 23 '>Ph 1 'ilxe2 24 h3, with a substantial
advantage to White.
18 'ilcl
There is nothing more for the queen to do on a4.

18
l:f8
After this, the black pieces come under pressure. Evidently 18

l: c8 (7) was more exact, trying to obtain counterplay along
the c-file.
7
w
From diagram 7, here are some important recent examples :

Tisdall-Thorsteins, Reykjavik 1 989, went 19 de be ( 1 9 . . . l: f8?
20 cb ll:lc4 21 g 1 ) 20 l: d 1 'ilxe5 21 'ila4 l: f8 22 l: d3 c4 23 l: f3
'lidS 24 l: xf8 + xf8 25 .i.xa7 ll:lc6 26 h3 'ile4 27 'ilb5! and
White had a considerable plus. However, 20 . . . 'llc4! is more
accurate : 21 .i.g3 'ilxg4 22 h3, with chances for both sides.
Lputian-Dzhandzhgava, Simferopol 1 988, went 19 'ildl l:d8
20 'ilcl (20 'ilc2 repeats moves) 20 . . . l: f8 (the position that has
now arisen is the same as in my game with Kasparov, except that
the white queen is on c l . The difference is not too important,
but . . . ) 21 h3 (this is a serious inaccuracy. Correct moves are 2 1
 g 1 and 2 1 g5) 2 1 . . . W"f7 2 2 W" e l .th6 2 3 ll:lg3 ll:lc4 24 e 6 W"g7
25 ll:le4 .te3 26 W"e2 b5 27 '>Pg 1 .txf2+ 28 ll:lxf2 cd (Black has
won his pawn back and seized the initiative; the game doesn't last
much longer) 29 cd W"xd4 30 :Z. c l g5 31 l: c2 ll:le3 32 l: d2 'ila l +
33 ll:ld l l:fl + 34 h2 W"e5+ 35 g3 'ile4 36 l: d8 + g7 37 'ilb2 +
r;Ph6 0- 1 .
In Douven-Ilincic, Alma-Ata 1 989, after 1 9 1W d 1 l:t d8, White

played 20 ll'el , an improvement on the previous example. There
followed 20 . . . .!tlc4 (a more precise method was 20 . . . cd 2 1 cd
-th6 22 g1, and only then 22 . . . .!t!c4) 2 1 g5 1We4 22 g1 1Wg4
23 .ltlg3 cd 24 cd 1Wxg5 25 1Wb4 l:t c8 26 .ltle4 1Wf4 27 1Wxe7 h8
28 We6 l:t f8 29 1W xc4 1Wxe4 30 : e 1 , with a clear advantage.
19 .ltlf4 is well answered by 19 . . . 11'f7 or 19 . . . 1Wc4+ 20 1Wd3
-th6. Another inadequate line is 19 1W d 2 : f8 20 1Wg5 1Wf7 2. 1
1Wh4 -th6! (but not 2 1 . . . .!tlc4? because of 22 g5!, as in Schneider
Ljubojevic, Pernik 1 988) 22 g1 .!tlc4!, and the initiative is with
Black. At move 20, the unfortunate 20 g1? led to a quick loss
in Schulz-Filipovic, Korimofen 1 989 : 20 . . . -*.xeS 21 l:t d 1
-txh2 + !
It only remains t o add that 18
1W c4 would be met b y 1 9 1We4
with favourable consolidation, for instance : 1 9 . . . l:t f8 20 g1 !
-th6 2 1 -th4!. Notice one peculiarity of the position resulting
from 1 7 . . . b6: Black now constantly has to reckon with the
exchange d4xc5.
.
19
g1
1Wc4
The bishop cannot now break out onto the open board: 1 9 . . .
-th6 20 h4 1Wf7 2 1 .ltlg3, or 1 9 . . . .!t!c4 20 h4.
20
1Wd2!
White continues with his plan of restricting the bishop's mobility.

After 20 1We4, he would have to reckon with 20 . . . -th6 and
especially with 20 . . . .!tlc6!?, threatening to capture on e5; naturally,
2 1 1Wxc6 1W xe2 cannot arouse White's enthusiasm.
20
1We6
20 . . . 1Wf7 is not good, since after 21 .ltlg3 everything fits together

for White : his knight is transferred to e4, and his queen to e2.
Nor does 20 . . . -th6 21 1Wxh6 1W xe2 rid black of his worries, on
account of 22 1We3 1Wxg4 23 de be (23 . . . .!tlc4 24 1Wd4) 24 1Wxc5.
21
22
h3
ll'gS! (8)
.!tlc4
A critical moment. In addition to .!tle2-f4, White will now be

threatening -tf2-h4 in some variations.
22
b6
22 . . . -tf6 is no good, if only because of 23 ef (23 .ltlf4 is also

strong) 23 . . . ef 24 .!tlf4.
11'7
23
ll'cl
To obtain counterplay Black had to opt for 23 . . . b5, intending
8
B
(for example) 24 ll:lf4 'flf7 25 ll:ld3 b4 !? even though 25 .tg3

maintains a plus for White. 23 .. . 11d5 would not work in view
of 24 'flc2!, while after 23 . . . h5 the queen would return to g5.
-
24
.tg3
g5
This move was roundly condemned by the commentators. But

I do not recall that any serious alternative was suggested to
11d5 (for example) has been
give Black counterplay. 24
recommended, but then 25 ll:lf4 'fle4 26 ll:le6 would put Black in
a very dangerous position. There are three variations - let us look
at them :
(a) 26
.l:l. c8 27 'ilb 1 ! 'fle3 + 28 .tf2 'flxc3 + 29 'flxg6 'flxa 1 +
30 h2, with unavoidable mate.
(b) 26
ll:le3 27 11d2 cd 28 cd .l:l. c8 29 .l:l. e 1 .l:l. c2 30 .l:l. xe3 'flc6
3 1 d5, and Black is lost. He could have tried driving the queen to
e2, with 27 . . . ll:lc4 28 'fle 1 ll:le3 29 'fle2 cd. Then 30 cd .l:l. c8 would
give Black strong counterplay, but 30 ll:lxf8 .txf8 (30 . . . d3 3 1
11f2) 3 1 'flf3! 'ild3 32 cd 'ilxd4 33 .l:l. e 1 clears up the situation
completely - since 33 . . . ll:lc2+ 34 .tf2 ll:lxe 1 is refuted by the
intermediate 35 'ilb3 + .
(c) 2 6
cd 27 ll:lxf8 ltle3 2 8 11d2 d e 2 9 'fle2 .txf8 3 0 'flf3, with
a big advantage.
25 'flc2
11d5

. .
. . .
26
27
.tf2
ll:lg3
b5
.1:1. 7
Forced. If 27 . . . b4, then 28 ll:lf5 is unpleasant, while 28 . . . .l:l. f7

is met by 29 e6 'flxe6 .l:l. e 1 11d7 3 1 cb.
28
.l:l.el
I f White had wanted, he could have put a stop to Black's
queenside counterplay with 28 : c t .
28
29
1Wg6
b4
f8
29 . . . be loses immediately to 30 o!Llf5 f8 3 1 e6 .l:l. xf5 32 gf

o!Lld6 33 de.
30
o!Lle4
30 o!Llf5 is also strong: 30 . . . e6 3 1 o!Llxh6 .l:l. f4 32 1W xg5 , or 32

1Wh7.
.l:l. xfl
30
The exchange sacrifice is incapable of improving matters. All it

demands from White is a modicum of accuracy.
31
32
33
34
35
36
37
39
xf2
'tiffS+
1Wc8+
1W xc5
g1
o!Llg3
o!Llf5
.l:l. c 1
1-0
be
g8
h7
1Wf7+
c2
.i.f8
g8
Black's pieces on the kingside never succeeded in breaking free.

As this game was played in the World Cup in the French
town of Belfort, someone jokingly called 1 7 1Wa4 the Belfort
variation . . . .
I might add that this game was declared the best game played
in the first half of 1 988 (Informator, vol. 45). Also, it proved to be
our last decisive encounter until the following (already our fifth)
World Championship match.
Game No. 3
Karpov-Timman
Rotterdam 1989
1 d4 o!Llf6 2 c4 g6 3 o!Llc3 d5 4 cd o!Llxd5 5 e4 o!Llxc3 6 be J.. g7 7
J.. c4 c5 8 o!Lle2 o!Llc6 9 J.. e3 0-0 10 o-o J.. g4 1 1 f3 o!Lla5 1 2
J.. xf7 + .l:l.xf7 13 fg l:Xfl+ 14 xfl 1Wd6 1 5 e 5 1Wd5 16 J.. f2
.l:l. d8
In the fifth game in Seville, Kasparov chose 1 6 . . . .l:l. f8 (see
Game No. 1 ). As we know, the rook move to d8 was first employed
in the 7th game of the match.
17
1Wc2
In the first game with this line, I chose 17 'tlfe l (see Game No.
2); in Belfort 1 988, I preferred 1 7 'ilt'a4 (Game No. 2).
17
'ilt'c4
In Griinberg-Ilincic, Prague 1 989, an equal game resulted from
1 7 . . . .l:t c8 1 8 illf4 'ilt'f7 1 9 'ilt'e4 J.h6 20 g3 .l:t f8 2 1 >g2 J.xf4
22 gf 'ilt'xf4 23 'ilt'xf4 .l:t xf4 24 >g3 .l:t f8.
18
'it' b2
But not 1 8 g5? on account of 1 8 . . . cd. If 1 8 'ilt'e4, then 1 8 . . .

.l:t d5!, threatening 1 9 . . . .l:t xe5, is quite good for Black.
J.h6
18
In Ljubojevic-Timman, Linares 1 989, Black moved his rook
again : 18
.l:t f8. There followed 19 >g1 J.h6 20 .l:t dl ! 'ilt'a4 2 1
.!:te l cd (better was 2 1 . . . illc4 2 2 'ilt'b3 'ilt'xb3 2 3 ab illb2) 2 2 illxd4
'ilt'c4 23 h3 b6 24 illf 3 .l:t d8 25 .td4 (25 g5! is also playable) 25
. . . J.f4 26 >f2 1Fd5 27 'it' b l ill c4 28 'ilt'e4 ill b2 29 'ilt'c2 'it'b5 30
.l:t b l illd 3 + 3 1 >fl 'ilt'c4 32 'ilt'e2 b5 33 ille l ! illx e5 34 .l:t xb5 'ilt'f7!
35 >g1 illc 6 36 illf 3 illxd4 37 cd .l:t c8 (but not 37 . . . J.d6 38
illg 5 'ilt'f4 39 'ilt'e6+ >g7 40 'ilt'xe7 + with advantage, or 39 . . .
>f8? 40 .l:t f5 + and wins) 38 .l:t c5 .l:t b8, and the players decided
to draw by repetition. It remains to be noted that at move 28,
White could have kept the initiative with 28 .l:t e2!
After 18 . . . .l:t f8 19 >gl J.h6 20 .l:t d 1 , an innovation, 20
'ilt'e6,
was played in L. Hansen-Jasnikowski, Warsaw 1 990. (Instead, 20
. . . 'ilt'f7 2 1 .!Llg3! illc4 22 'ilt'e2 'ilt'xf2+ 23 'ilt'xf2 J.e3 24 ille4! .l:t f4
25 .l:t d3 is hardly good for Black.) However, after 2 1 h3 .!Llc4 22
'ilt'xb7 .te3 (22 . . . ille 3 23 .!:ta l 'ilt'f7 24 J.xe3 J.xe3 + 25 >h l
'ilt'f2 26 'it'b5 cd 27 illx d4!) 23 J.xe3 illx e3 24 .l:t d3 'ilt'f7 (or 24 . . .
.l:t fl + 25 >h2 ill xg4+ 26 hg 'ilt'xg4 27 .l:t f3 'ilt'h5 + 28 .l:t h3 'ilt'xe2
29 'ilt'xe7 .l:t f7 30 'ilt'xc5 and wins - Hansen) 25 'ilt'f3! 'ilt'xf3 26 gf
.l:t xf3 27 de! White had a clear endgame advantage.
...
19
9
B
h4 (9)
This position first occurred in Karpov-Kasparov, Amsterdam

1 988, which proceeded as follows : 19
11f7 20 gl .J:I. f8 21 .!Dg3
(2 1 .ig3 is met by 21 . . . .ie3 + 22 h2 11e6 23 g5 11g4) 21 . . .
.!Dc4 22 'ife2 11xf2 + ! (other continuations lead to advantage for
White : 22 .id2 23 .!De4, or 22 . . . cd 23 cd 11xf2+ 24 11xf2 .ie3
25 11xe3 .!Dxe3 26 .J:I. c l ) 23 11xf2 .ie3 24 11xe3 (24 .J:I. fl ? .J:I. xf2 25
.J:I. xf2 cd) 24 . . . .!Dxe3 25 de .J:I. c8 (but not 25 . . . .!Dxg4 26 .J:I. b l ,
when White i s already better) 2 6 .J:I. b 1 .J:I.xc5 2 7 .J:I.xb7 .!Dxg4 28
.J:I. b4 (White may also play 28 : xe7 : xc3 29 .!Dfl : a3 30 e6
.J:I. xa2, which quickly draws) 28 . . . h5 29 .!De4 .J:I. xe5 30 g3 (30 fl
.!De3 + 3 1 e2 .!Dd5 32 .J:I. b 8 + is simpler) 30 . . . f7 3 1 g2
e6 32 f3 (a more accurate choice is 32 .!Dg5+ f6 33 .J:I.f4+
.J:I.f5 34 .J:I.e4 .J:I.f2 + 3 5 g1 .J:I.xa2 36 .:t f4+ e5 37 .J:I.e4+ d6
38 .J:I. e6+ d7 39 .J:I. xg6, and a draw can be agreed) 32 . . . f5!.
Black now has slightly the better endgame, not that there is much
danger to White. After a further two dozen moves, we agreed a
draw.
The question arises whether the strange retreat with the queen
to f7 can really solve all Black's problems. It turns out that this
is not quite the case. In Vyzhmanavin-Ernst, Stockholm 1 990,
Black was dealt an annihilating blow, radically altering the
assessment of the variation. The game went 19 . . . 11f7 20 g5!
(White isn't afraid of the pin on his bishop. The main thing is to
shut the opposing bishop out of play) 20 . . . .!Dc4 2 1 e6! (this is
just the intermediate move that constitutes the refutation of Black's
defence. White deflects the queen, now allowing his opponent to
double on the f-file, and only afterwards picks up the b-pawn) 2 1
. . . 11f5 2 2 .!Dg3 11xe6 2 3 11xb7 .J:I. f8 24 .J:I.e 1 ! (now Black can't
avoid loss of material) 24 . . . .!De3 + 25 g l .J:I.xf2 26 gh cd 27 cd
.:t f8 28 11b2 lljc4 29 .J:I.xe6 1 -0.
Perhaps Timman had a foreboding of such misfortunes, and
therefore avoided the move that justified itself in Amsterdam ( 1 9
. . . 11f7), preferring t o occupy the f-file with his rook.
.
19
20
.1:1.8
g5!
The stock manoeuvre; here too it guarantees White a plus.

20
11d3
21
'ilbl!
Gaining a couple of tempi for bringing the queen to the defence.
A mistake would be 2 1 gh? lbc4 22 1i' c l 1Wf5 23 it'e l lbe3 + 24
g l lbc2.
21
22
23
it'el
gl
1i' e3
R.g7
But not 23 lbc l ? cd and everything is fine for Black, as after 24

1t'xe3de, he wins.
1t' e4 (10)
23
10
w
24
lLig3!
White returns the pawn but seizes all the key squares with his
pieces. A familiar precept!
24
25
1t'xh4
lbe4
25 lLif5 is also strong : 25 .. . 1t'xg5 26 lLixg7 irg4 (26 . . . xg7

27 R.h4 1i'g4 28 R.xe7, etc.) 27 d5 xg7 28 R.xc5 1t'd7 29 1i'e4,
and Black is in a bad way.
l:txf2
25
An exchange sacrifice (again not entirely voluntary!) analogous

to the one that occurred in Karpov-Kasparov at Belfort. 25 . . .
1t'g4 does not help either : 26 lLixc5 b6 27 1i'e4 1t'xg5 28 1i'd5+
l:t f7 29 lbe6, and White should win.
26
27
lLixf2
l:tdl !
cd
d3
On 27 . . . lLic6, I had in mind 28 cd lbxd4 29 1i'e3 lbc6 30 1t'b3+

f8 3 1 1i'xb7 1t'c4 32 1i'c8 + f7 33 lbg4!. Nor is 27 . . . de 28
1i' xc3 lbc6 29 1t'b3 + any better for Black.
lLic6
28 1We3!
28 . . . lLic4 is well answered by 29 1W d4.
28 eville Variation
29
30
31
32
33
If 33
. .
tl:lxd3
'lrf3!
e6
tl:lf4
tl:ld5!
'lra4
'lraS
tl:ld8
.i.eS
'IreS+
. tl:lxe6, then 34 l:t fl ! is decisive.

34
h1
1-0
There is no defence against the threats of 35 tl:lf6+ and 35 : fl.

Game No. 4
Naumkin-Neverov
Moscow 1989
1 d4 tl:lf6 2 c4 g6 3 tl:lc3 d5 4 cd tl:lxdS 5 e4 tl:lxc3 6 be .i.g7 7
.i.c4 c5 8 tl:le2 tl:lc6 9 .i.e3 0-0 10 0-0 .i.g4 11 f3 tl:laS 12
.t.xf7+ .:.x7 13 fg J: xfl + 1 4 xfl 'ir d6
15
g1
In the 5th and 7th match games in Seville I continued 1 5 e5,

and the current state of theory on that line is given in the notes
to the first three games of this book. In the 1 1 th match game,
however, I played the king to g1 here. White is prepared to return
the pawn in order to obtain a sturdy centre. But before going on
to a detailed discussion of the prospects for either side, it is worth
mentioning one other possibility : 15 'lra4!?
This occurred in Zakharov-Henkin, Voronezh 1 989, where
Black replied 1 5 . . . 'lrxh2! (11) .
ll
w
A bold piece sacrifice. Instead, the timid 1 5 . . . b6 1 6 g 1 cd

1 7 cd 'lre6 1 8 d5! 'lrxg4 1 9 :tel , followed by 20 h3, is in White's
favour.
The game proceeded : 1 6 'ifxa5 l:l. f8 + 1 7 e1 (but not 1 7 J.. f2?
on account of 1 7 . . . 'if h4! 1 8 g3 'ii' h 1 + 1 9 g1 'ifh2, and wins)
1 7 . . . 'ii' h 1 + 1 8 g 1 ! (on 1 8 .!.g1 ?!, Black has the very strong 1 8
. . . ..th6! 1 9 'ifxc5 'if xg2, threatening 20 . . . 'iffl mate) 1 8 . . . 'ifxg2
(but here, 1 8 . . . J.. h 6 comes too late : 1 9 d2 l:l. f2 + 20 d3
c4+ 21 .;;.xc4 J.. x e3 22 'ii' d 5+ .;;.g 7 23 l:l. e 1 , with the better
endgame chances for White; if he prefers, he can give perpetual
check by 23 f3 'ifxa 1 24 'ife4+) 19 'ii' b5 ..th6 (after 19 . . . cd 20
cd 'if xe4 2 1 'ife2 and 22 l:l. d 1 , White consolidates; but now, a
highly unusual distribution of forces comes into being) 20 'ife2
'ifg3+ 21 d2 l:l. f2 22 J.. x h6 l:l. xe2 + 23 xe2 'ifh3! 24 J.. e 3!
'ifxg4 25 l:l.fl ! h6! (a very strong move according to Henkin, who
gives the following variations : 25 . . . cd 26 J.. h 6! de+ 27 c1
'ifc8 28 d4, threatening e6 and l:l.f8 +; 25 . . . h5 26 f4,
followed by l:l. g 1 ; or 25 . . . g7 26 f4 g5 27 l:l. g 1 'ifh4 28
l:l. xg5 + ! . Now the game heads towards a draw) 26 J.. x h6 'ifxe4
27 l:l. f8 + h7 28 .!.g5 'ife6 29 a4 cd 30 cd 'ifa2+ 3 1 .;;.el 'ifxa4
32 lH7 + .;;.g 8 33 .J:I. xe7 'ii' b 4+ 34 f2 a5 35 J.. f6 'ii' b 6! 36 J.. e 5
a4 37 f4 'ii' b 2+ (Black loses after 37 . . . f8 38 J.. g 7+ !, or 37
. . . a3 38 e6!) 38 g1 ! 'ifc1 + 39 h2 'ii' b 2+ 40 g1 'ii' b 1 + 41
.;;.h 2 'ii' b2 + t-t.
So the extravagant 1 5 'ifa4 is not dangerous to Black. It must
be acknowledged that 1 5 e5, which we have already examined, is
the most precise; White can move his king if the need arises.
Nonetheless, 1 5 .;;.g 1 should also be studied.
15
'ife6
In Polajzer-Anka, Dortmund 1 988, Black played 15
l:l.d8.
It isn't clear that the rook on d8 is useful to Black after 16 'ii' d 3
'ife6 1 7 g5!. However, White replied 16 'if a4 'ifa6 1 7 .J:I.e 1 'ii' d 3 1 8
J.. f 2 c4 1 9 'ii' b5 l:l.f8 20 'ii' b l 'ii' d2 2 1 'ii' b3 b 5 2 2 'ii' x b5 e3 23
h3, and now, in Gutman's view, 23 . . . J.. h 6! would have given
Black a powerful initiative.
16 'ii' d3
'ifc4
So far, the play has followed the 1 1 th game in Seville, and the
text move is better than the capture on g4, which occurred (with
a transposition of moves and the insertion of the pawn exchange
on d4) in the 9th match game. In Chernin-Malishauskas, Lvov
1987 (played while the Seville match was still in progress), Black
didn't hurry to exchange queens, preferring 16
cd 17 cd l:l. d8.
After 18 g5 c4 1 9 J.. f2 b5 20 a4 (20 h3 was worth considering)
.
20 . . . ba 2 1 lllf4 'flf7 22 lll d 5 .l: f8 23 i.g3 lllb 6 (but not 23 . . .

lllb 2 24 'flc3 a3 25 h3, followed by i.e5, with advantage) 24 h4
(24 lll x b6 'ilb3!) 24 . . . lllxd5 25 ed 'ilxd5, the chances were equal.
Instead of 1 8 g5 White can play 18 h3, but after the queen exchange
Black's chances, once again, are no worse : 1 8 . . . 'ilc4 19 'ilxc4
lllxc4 20 i.f2 e5 2 1 d5 i.h6 22 h4 (22 a4 llld 6!) 22 . . . i.d2 23
a4 llld 6!. Note that variations involving the pawn exchange in the
centre ( . . . c5xd4, c3xd4) will be examined more closely in Game
No. 5.
17
'ilxc4
It hardly pays White to avoid the queen exchange : 1 7 'ild2 cd
1 8 cd 'fla6! or 1 7 . . . 'fle6.
lllxc4
17
18
i.f2
Vasser Seirawan, who has also incorporated this variation in
his repertoire, twice played 18 i.gS (12) here. Both games are
worth examining.
12
B
Seirawan-Lputian, St John 1 988, continued 18

h6 1 9 i.xe7
cd (Black can win a piece with 1 9 . . . .l: e 8 20 .i.xc5 b6, but after
2 1 i.xb6 ab 22 e5, White obtains more than enough pawns for
it. Furthermore, once his king is centralised and the knight jumps
to e4 via g3, the white pawn mass can become highly mobile) 20
cd .l: e8 21 .l: c 1 lll a 5 (2 1 . . . .l: xe7 22 ::txc4 .l: xe4 23 f2 .l: xg4
24 J:t c8 + h7 25 .l: c7 is no better for Black, but 2 1 . . . b5 was
worth considering) 22 .l: c7 lllc 6 23 i.c5 J:t xe4 24 ci>f2 J.xd4+
(in the event of 24 . . . lllxd4, Lputian gives 25 .1: xg7 +! xg7 26
i.xd4+ f7 27 f3 J:t e6 28 i.xa7 J:t a6 29 J.e3 J:t xa2 30
i.xh6 b5 3 1 lllc 1, with advantage) 25 lllxd4 lllx d4 26 .l: xb7 lllc 6
. .
27 h3 :la4 28 a3. White has an extra pawn, which he eventually
exploited to win (though it took him 30 moves!).
Seirawan-Hort, Lugano 1 988, went 18
cd 19 cd eS 20 :lc1
(20 d5 h6 2 1 .i.cl d6 22 g3 : c8 23 .i.e3 :lc3 24 f2 .i.f6!
2S a4 .i.gS led to equality in Dlugy-Nikoloff, Toronto 1 989;
Black can also play the immediate 20 . . . tLld6 21 g3 : c8 22
.i.e3, with adequate counterplay) 20 . . . bS 2 1 de .i.xeS 22 :ld 1
: c8 (22 . . . : e8 is more accurate and gives equality} 23 .i.f4 !
.i. g 7 2 4 :ldS a 6 2S f2 :le8 2 6 f3 f7 27 h 4 e S + 2 8 .i.xeS
: xeS 29 : d3 b4 30 f4. Here again, White has an extra pawn,
but this time Black managed to hold out.
Interestingly, his loss to Seirawan made such a strong impression
on Lputian that he took the first opportunity to play this variation
with White. In Lputian-Hansen, Dortmund 1 988, Black played
(from diagram 1 2} the immediate 18
e5. There followed 19 d5
b5 ( 1 9 . . . h6 20 .i.cl is sounder for Black} 20 : b1 :b8 21 f2
aS 22 thc 1 h6 (22 . . . a3 23 :lb3 b4 24 cb cb 25 d6) 23 .i.e3
xe3 (in Lputian's view, 23 . . . tLld6 24 f3 :lf8 + 25 e2 c4 26
e1 xe4 27 :lxbS xc3 28 :lxaS :ld8 would have given
equality; but not 23 . . . :lf8 + 24 e2 xe3 2S xe3 :lfl 26
:la 1 ! and 27 b3!} 24 xe3 c4 2S e2 .i.f8 26 tZ'lg1 .i.cS + 27
e2 .i.xg 1 28 :lxg 1 f7 29 a3 rj;e7 (29 . . . b4 would have
retained some saving chances} 30 :lb 1 , and White won the rook
ending. After the game, Lputian explained that at move 2 1 , the
correct course was 21 a4! b4 22 cb cb (or 22 . . . :lxb4 23 tZ'lc3
d6 24 :le 1 , preparing 2S .i.e7) 23 c1 .i.f8 24 b3 d6 2S
d2 b3 26 .i.e3 a6 27 fl! :lb4 (28 . . . b2 29 e2) 28 .i.c5
:lxa4 29 .i.xd6 .i.xd6 30 :lxb3, and White has the better chances.
.
cd
18
Black can play 1 8 . . . eS at once, without opening the c-file.

After 1 9 dS ( 1 9 de :ld8 20 gS .i.f8} 1 9 . . . b6 20 gS :lf8, he has
a perfectly secure fortress.
19
cd
e5
After 19 . . . b6 or 19 . . . bS, White plays 20 :lbl! with the better

chances.
20
d5 (13)
20 :lcl d6 2 1 de .i.xeS is not dangerous for Black.

20
d6
A refinement on the 1 1 th game in Seville. But before going any

further, let us recall how that game (which played an important
13
B
part in the match) proceeded. It might, incidentally, have been

worth including it (accompanied as it is by some new annotations)
among the 'basic' games in this volume. But we will not break
our rule: only games played in 1 988 or later are to be specially
numbered.
20 . .. .t..h6 21 h4 .t..d2 22 'fJ.d1 .t..a5
If 22 . . . b5, White has 23 lt! c 1 !, heading for b3. Gutman quotes
the following analysis by H. Wirthensohn : 23 . . . aS (23 . . . .t..x c1
24 'fJ.xc1 lt!d6 25 'fJ.e 1 a6 26 .t..g 3 'fJ.e8 27 <;Pf2) 24 lt!b3 .t..b4 (24
. . . .t..c 3 25 d6 a4 26 d7 rJ.d8 27 lt!c5 lt!b6 28 rJ.d6) 25 .t..c 5 .t..x c5
26 lt!xc5 'fJ.c8 (26 . . . lt!d6 27 lt!d7 lt!xe4 28 lt!xe5 lt!d6 29 'fJ.c 1 )
2 7 lt!b7 'fJ.c7 2 8 d 6 'fJ.d7 29 'fJ.d5 lt!b6 3 0 lt!c5, and White i s on
top.
23 'fJ.c1 b5
Here 23 . . . lt!d6 is no good : 24 lt!g3 .t..b 6 25 J.xb6 ab 26 'fJ.c7
'fJ.a4 27 'fJ.e7 lt!xe4 28 lt!xe4 'fJ.xe4 29 d6 f8 30 'fJ.xh7 'iPe8 3 1
h5, and Black is doomed.
24 'fJ.c2 ltld6 25 ltlg3 ltlc4 26 ltlf1 ltld6 27 ltlg3 ltlc4 28 g5 /7
White has consolidated his advantage. If 28 . . . a6 (28 . . . .t..b 6
29 a4 lt!a3 30 'fJ.b2 ba 3 1 'fJ.a2), there follows 29 lt!fl lt!d6 30 'fJ.c6
lt!xe4 3 1 lt!g3! lt!c3 32 d6! f7 33 d7 'fJ.d8 34 .t..c 5 'fJ.xd7 35
'fJ.f6 + g8 36 h5 gh 37 lt!xh5 'fJ.d8 38 'fJ.xa6 .t..c 7 39 lt!f6+, and
Black is in a bad way.
29 ltl/1 ltld6 30 ltlg3 ltlc4 31 /1 e7
Korchnoi suggested 31 . . . .t..b 6 32 .t.. x b6 ab 33 'iPe2 'fJ.a4, but
a stratagem we have seen before - 32 a4! lt!a3 33 'fJ. b2 ba 34
'fJ.a2 - gives White the advantage.
32 .t..c5+ <;Pf7 (14)
14
w
33 J:.f2+
A critical moment in the game. After the obvious 33 e2, Black
would scarcely be able to save himself: 33 . . . i.. b 6 34 a4 i.. x c5
35 ab J:. b8 36 d3! lhb5 (36 . . . a6 37 xc4 ab + 38 b3 i.. d6
39 J:. c6 e7 40 e2 J:. a8 41 c3 J:. a3 + 42 b2, or 40 . . b4
4 1 c1, with a won position) 37 xc4 J:.b7 38 lL!fl i..d 6 (38 . . .
a5 3 9 d2, o r 38 . . . i.. b4 3 9 e3) 39 d2, etc.
33 . . . gl 34 J:.f6 i..b6 35 J:.c6??
By retreating his bishop to f2, White would preserve the better
chances. Instead of this, I commit a fatal blunder. The sad fact is
that the possibility of the black knight jumping to the edge of the
board has escaped my attention.
35 . . . a5 36 i..xb6 xc6 37 i..cl J:.f8+ 38 e2 J:.fl 39
i..d6 J:.dl 40 i..c5 a5 41 !1 .:tel 42 i..d6 J:.c2+ 43 d3
J:.xa2 44 e3 fl 45 g4 c4 46 xe5+ xe5 47 i..xe5 b4
48 i..f6 b3 49 e5 J:.xg2 50 e6 + f8! 0-1.
Now at last we return to the game Naumkin-Neverov, which,
as it happens, was decided very quickly.
i..h 6
21
g3
.
22
i..cS
White proceeds ineffectively and is soon in a difficult position.

The immediate 22 h4 was better.
22
c4
23
24
25
26
h4
i..f 2
e2
J:.dl
J:. c8!
i..f4
i.. d2
aS!
White's loss of time in moving his bishop about begins to tell:

Black creates a passed pawn, which decides the issue.
27 gS bS 28 ltlg3 b4 29 lll fl J.c3 30 lll h 2 a4 31 d6 b3 32 ab ab
33 ltlg4 b2 34 d7 .IUS 35 :n <t;g7 36 ltlf6 lll d2 37 ltle8 + <t;f7
o-1
So 1 2 J.xf7 + doesn't guarantee victory. One or two inaccur

acies, and Black may assume a decisive initiative.
Game No. 5
Seirawan-Popovic
Manila 1990
1 d4 ltlf6 2 c4 g6 3 ltlc3 dS 4 cd lll xdS 5 e4 lll xc3 6 be J.g7 7
J.c4 cS 8 lll e2 lll c6 9 J.e3 o-o 10 0-0 J.g4 1 1 f3 ltlaS 12
J.xf7 + ll xf7 13 fg ll xfl+
14
xfl
cd
15
cd (15)
15
B
Curiously enough, this precise position arose as long ago as

1 955, in the game Spassky-Korchnoi from the 22nd USSR
Championship. Play continued : 15 ... 1i'd7 16 h3 1i'e6 17 1i'd3
1i'c4 18 1i'd2 (or 1 8 1i'xc4 lll xc4 19 J.g5 e6 20 ll d 1 b5, and despite
White's extra pawn and the queen exchange, the initiative is with
Black; Panteleyev-Prakhov, Bulgaria 1 970. An improvement was
20 .l:l bl !) 1 8 . . . 1i'a6! 19 1i'c2 ltlc4 20 1rb3 <t;h8 21 <t; g 1 ltld2 (21
. . . .l:l f8 is stronger) 22 J.xd2 1i'xe2 23 .te3, with equal chances.
If Black wants, he can force a draw with 23 . . . .l:l c8 24 .l:lfl .l:l c2
25 ll f2 1i'el+ 26 ll fl 1i'e2.
The immediate pawn exchange on d4 at move 14 occurred only
once in the Seville match - in the ninth game. Let us take quite
a close look at it (recently some new thoughts about it have come
to mind!):
15 . . . flb6 16 g 1
With the queen on b6, 1 6 h3 can be strongly answered by 1 6
. . . c4 1 7 ..t.f2 flb2! 1 8 gl l:t f8 1 9 l:t b l flxa2 20 l:t xb7 .th6!
with the initiative.
16 ... fle6
With the white king on g 1 , the line just mentioned no longer
gives Black anything : 16 . . . c4 1 7 .tf2 flb2 1 8 f4! l:t f8 1 9
d3.
17 fld3!
White returns the pawn while retaining all the positional trumps.
The position had first been seen (it arose by transposition) in
Alfeyevsky-Werner, corr. 1 984. That game went 17 g3 l:t d8 1 8
l:t c l ( 1 8 l:t b l .txd4 1 9 .txd4 c6 20 e2 xd4 2 1 xd4 flc4)
1 8 . . . flb6 19 e2 c6 20 l:t b l xd4! 2 1 l:t xb6 f3 + 22 f2
l:t xd l 23 l:t xb7 xh2 24 g4 g4+ 25 f3 h2 + , with a peaceful
conclusion by perpetual check.
1 7 . . . flxg4
After this, I succeed in obtaining a substantial plus. After 1 7 . . .
l:t d8 1 8 g5 c4 1 9 .tf2 b5, the play would transpose into
Chernin-Malishauskas which we have examined before (see page
3 1 ). (I should point out that the ideas and variations in Games
1 -5 have much in common, while sometimes being distinguished
by small nuances; I hope the reader will not get tangled up in the
'undergrowth'.) Instead of the capture on g4, a better idea is 1 7
. . . flc4, aiming to exchange o ff the strongest pieces. This idea was
employed in game 1 1 of the Seville match, with the difference that
the exchange on d4 took place later (see notes to Game No. 4).
White may avoid the queen exchange by retreating with 1 8 fld2
(exploiting the absence of the pawn from c3 and thus gaining a
tempo). After 1 8 . . . fla6 19 flc2 c4 ( 1 9 . . . flc4 is well answered
by 20 :t e l ! - emphasising once again that the removal of the c
pawns is favourable to White) 20 .tf2 .tf8 (20 . . . .th6 21 h4)
21 g5, and Black's position is devoid of counterplay.
18 l:t/1 l:tc8
The alternative 1 8 . . . l:t f8 is worse : 1 9 :t xf8 + .txf8 20 d5 b6
2 1 d4!
19 h3! fld7 20 d5 c4 21 ..t.d4
More precise than 2 1 d4 xe3 22 flxe3 l:t c4 23 e6 flxe6
24 de .td4 25 flxd4 l:t xd4 26 l:t f7 l:t xe4 27 l:t xe7 b5, with
equality; but 21 .txa7 b6 22 ..t.b8!? was worth considering.
21 . e5
If 2 1 . . . lbe5, then 22 ..txe5 .txe5 23 lbd4 .txd4+ 24 -.xd4
b6 25 e5 -.c7 26 -.g4! is not bad for White.
22 de -.xe6 23 .t xg7 r;J;xg7 24 lb/4 -.d6 25 -.c3+ (16)
.
16
B
25 ... h6
The only move. There would be a striking finish after 25 . . .

-.es 26 lbe6 + ! g8 (26 . . . r;J;h6 27 'if c 1 + g5 28 l:l.f5) 27 -.xc4 !
l:l. xc4 28 l:l. f8 mate. Another line that fails is 25 . . . r;J;g8 26 lbd5
lbb6 27 lbf6+ h8 28 -.b2 -.c5 + 29 h2 -.c3 30 lbe8! lbd7
(30 . . . g8 31 -.bs l:l.c5 32 lbf6+!) 31 -.xc3 l:l.xc3 32 l:l.f7 l:l. d3
.(32 . . . l:l. c8 33 lbd6) 33 e5 r;J;g8 34 e6, and White wins.
26 lbd5?
Black's king is in quite a dangerous position, and White should
have tried to approach it from the other direction with 26 lbd3!
The threat of lbd3-f2-g4+ is then fairly serious (26 . . . lbb6 27
'ifd2+ g7 28 'ifb2+ r;J;g8 29 -.b3 + h8 30 l:l.f7, with a
winning position).
26 . . . -.e5 27 Vd3
White could have maintained the tension with 27 'ife l . Many
annotators recommended 27 -.b4 instead, but this is not dangerous
for Black : 27 . . . 'ifxe4! 28 lbf6 aS! 29 -.bs (29 'ifc3 -.e3 +, or 29
'if a4 'ifd4+ 30 r;J;h 1 l:l.f8 3 1 lbg4+ g7, with advantage to
Black) 29 . . . 'ifd4+ 30 .l:tf2 (in the event of 30 h 1 lbe3 3 1 'if xb7
lbxfl 32 lbg4+, the black king fearlessly marches forward : 32 . . .
r;J;g5 33 -.e7 + r;J;f4 34 -.f7 + r;J;g3 35 -.xn h5 36 -.f3 + h4
37 g3 + xh3 38 lbf2 + -.xf2! 39 -.xf2 l:l. c l + , and wins) 30 . . .
-.at+ 3 1 l:l. fl -.d4+ 32 l:l.f2 -.a l +, and the contest could end in
perpetual check.
Seville Variation 3 7
27 . . . g7 28 !i:Jf6 11d6 29 11 c3 11e5 3 0 11d3 11d6 31 11c3 11e5
32 1ib3 l:l. c7 33 11 d3 l:l./7 34 11 xc4 l:l. x/6 35 l:l. dl b5
The position is drawn, but Black is playing somewhat recklessly.
A simpler method is 35 . . . l:l. e6 36 l:t d7 + l:l. e7 37 l:l. xe7 + 11xe7
38 11d4+ f7 39 11xa7 'lrxe4.
36 l:l. d7+ h6 37 11e2 11 c5 + 38 h2 11e5+ 39 g3 11 c3 40
g2 11 c4 41 11e3 + g5 42 l:l. d2 11/1 + 43 h2 11/3
The sealed move. Once again, just as happened in game 7 of
the match, I persistently sought winning chances after resumption,
and reached an endgame a pawn up - but alas, a drawn result
was unavoidable. We will follow this interesting game to the end.
44 11d4 l:l. e6 45 e5 11/5 46 l:l. e2 a5 47 11d5 b4 48 11xa5 11d3 49
l:l.g2 11d4 50 11a8 11xe5 51 11/8+ g6 52 11xb4 h5 53 h4 gh 54
11xh4 l:l. d6 55 11 c4 l:l. d4 56 11 c6+ g7 5 7 11b7+ h6 58 11 c6+
g7 59 l:l. c2 l:l. h4+ 60 g2 11e4+ 61 11xe4 l:l. xe4 62 l:l. c7+
g6 63 l:l. a7 l:l. e3 64 h3 l:tc3 65 l:ta8 l:l. c4 66 a4 g5 67 a5
l:l. a4 68 a6 <i;h6 69 g2 l:l. a3 70 /2 g7 t- t .
15
eS!?
This move had first occurred in the game Kir. Georgiev

Ivanchuk, Reggio Emilia 1 989/90.
16
dS
The natural reaction - White creates a passed pawn. Despite

the pawn extra, the endgame arising from 16 de i.xe5 1 7 11 xd8+
l:l. xd8 holds no danger for Black. Seirawan-H. Olafsson, Reykjavik
1 990, continued 1 8 l:l. c l !i:Jc6 19 g3 l:l. d3 20 i.f4 i.d4 21 !i:Jxd4
l:l. xd4 22 l:l. b l l:l. b4 23 l:l. xb4 !i:Jxb4 24 a4 !i:Jc6 25 e2 a6 26 i.d2
b5 27 ab ab 28 i.c3 rj;f7 29 <i;e3 e6. Subsequently Black tried
to make something of his passed b-pawn, but without success,
and a draw was agreed after another 25 moves.
Nor does White do any better with 16 l:l. cl 11d7 17 de 11xd 1 +
1 8 : xd 1 !i:Jc4 1 9 i.f2 i.xe5 20 l:l. d7 b 6 2 1 i.d4 i.xd4 22 !i:Jxd4
!i:Jd2+ 23 <i;e2 !i:Jxe4 24 !i:Jc6 a5 25 l:t b7 !i:Jc3 + 26 <i;d2 !i:Jxa2
27 : xb6 !i:Jb4, and the game is level; Schiissler-Kudrin, Saint
Martin 1 990.
!i:Jc4
16
17
i.f2
11 f6
1s
g1
: rs
19
11e1
The position is unclear after 19 i. x a 7 b6, or 19 i.c5 l:t c8 20

i.xa7 b6.
19
R.h6
Threatening . . . R.d2.
W'a6
20
lllg3
20 . . . W'xf2+ would be over-hasty : 2 1 W' xf2 R.e3 22 W' xe3,
followed by r:t cl.
21 <lo>h1
The threat was . . . r:t xf2, for example : 21 W'e2 r:t xf2! 22 <lo>xf2
R.e3 +; while if 2 1 lllfl , then 2 1 . . . lllb 2! is decisive.
21
W'a4
2 1 . . . W'a3 is also worth trying.
22
R.gl! (1 7)
17
B
The white queen's problems can be solved later. In Georgievlvanchuk, White played 22 "ife2, and there followed : 22
b6 23
h4 R.f4 24 lllfl (24 g5 llld 6 25 r:t e 1 was more precise) 24 . . . llld 6
25 l:l e 1 l:l c8 26 g3 r:t c2 27 W'f3 W'xa2 28 <lo>g1 R.h6 29 g5 R.g7
30 llle 3 r:t c7 3 1 lll g4 r:t f7 32 W'e3 W'c2 33 h5 lllc4 34 W'c1 W'xc1
35 r:t xcl gh 36 r:t xc4 hg 37 r:t c8 + R.f8 38 R.e l ! (lvanchuk gives
38 <lo>fl ! b5 39 <lo>e2 b4 40 R.c5 a5 41 r:t a8 b3 42 <lo>d2, with a
great deal of play left; or 38 . . . <lo>g7 39 <lo>e2 b5 40 r:t a8) 38 . . .
<lo>g7 39 R.c3 R.d6! 40 r:t c6 R.c5+ 41 <lo>g2 r:t f2 + 42 <lo> h 1 R.d4
43 R.b4 r:t f7! 44 r:t e6 r:t b7! 45 r:t c6 a5 46 d6? (lvanchuk considers
this the decisive mistake, although even after the more stubborn
46 R.a3 b5 47 d6 r:t f7! 48 r:t c7 R.b6 49 .D. c6 R.d8 50 R.b2 r:t d7
5 1 R.xe5 + <lo>f7, Black has a clear plus; not however 47 . . . R.b6??
48 r:t xb6!) 46 . . . ab! 47 r:t c7+ <lo>f8 48 r:t xb7 b3 49 r:t b8 + <lo>f7
50 d7 b2 5 1 r:t f8 + <lo>e6 52 d8 lll + <lo>e7 53 r:t fl xd8, and White
soon resigned.
. . .
One other game is worth mentioning. In Naumkin-Miralles,
Voskresensk 1 990, Black answered 22 11re2 with 22
.i.g5, and
only after 23 lOft did he play 23 . . . b6. (He could have brought
about simplifications with the bold 23 . . . l0d6! 24 .i.c5 l0xe4! 25
.i.xf8 11rd4 26 .i.c5 ! 11rxa 1 27 11r xe4 11r xfl + 30 .i.g1 .i.d8,
threatening . . . .i.b6.) Naumkin now seized the initiative: 24 h4
.i.e7 25 g5 .i.c5 (25 . . . l0d6 26 l0d2) 26 .i.g3 l0d6 27 l0d2 .i.d4
28 J: c 1 ! 1ixa2 29 J: c6! J: d8? (it was better to take the central
pawn : 29 . . . l0xe4 30 11rxe4 11r xd2 3 1 J: xg6+ hg 32 11r xg6 + h8
33 11rh6+, with a draw; sharp play would result from 3 1 h2
11re3 32 11rg4!?) 30 h2! aS (Naumkin gives these variations : 30
. . . 11r a 1 3 1 l0f3, or 30 . . . l0xe4 3 1 11rxe4 11rxd2 32 J: xg6+ hg 33
11rxg6+ ci>f8 34 11rf6 + ci>e8 35 11re6+ ci>f8 36 g6 .i.g 1 + 37
xg1 11rxd5 38 11rf6 + , with a substantial advantage to White) 3 1
h 5 11r a 1 32 l0f3 gh (nor can he save himself with 32 . . . 11rb2 33
l0xd4 11rxd4 34 11rg4 11rxe4 35 11re6+ .!On 36 J: c8!) 33 l0xd4 11rxd4
34 11rxh5 l0e8 35 g6 1 -0.

22
b6
Before playing the knight on d6, it is essential to cover the c5square.

23 11rc3
J: f7
23 . . . l0d6 is premature because of 24 11rxe5 .i.g7 25 11re6 + .
2 4 J: bl
.i.d2
24 . . . 11rxa2 is bad on account of 25 .1: b4, and the pawn on e5
falls. Seirawan recommends 24 . . . l0d2 25 J: b4 11r d 1 26 11r xe5
.i.e3 27 11re8 + with a draw, but instead White can maintain the
pressure with 25 .1: e 1 .
25 11rd3
After 25 11rb3 11r xb3 26 ab l0d6, White's extra pawn is of no
significance.
11rxa2
25
25 . . . .i.h6 (25 . . . .i.f4 26 l0e2 and l0c3) is answered by 26 a3,

and the a-pawn is invulnerable : 26 . . . l0xa3 27 : a t .
26
d6!
l0xd6
The only move. On 26 . . . J:t d7, Seirawan gives 27 11rd5 + ci>g7

28 J:t fl ! (28 11re6 J:t xd6 29 11re7+ ci>g8 30 J:t fl .i.f4) 28 . . . .i.f4
29 J:t xf4! ef 30 .i.d4+ h6 3 1 g5 mate. (Editor 's note - it appears
that Black can escape with 30 . . . f8. Therefore White should
prefer Stohl's 30 lt:lf5 + ! ! gf (30 . . . ci>f8 3 1 .!Oh6 g7 32 g5!) 3 1
i.d4 + ci>f8 3 2 11rxf5 + ci>g8 3 3 11r g5 + , forcing mate).
'irxb1
27 'irxd6
28 'irxd2
aS
28 . 'irb5 29 h3 'ird7 30 'irc3 'ire? 3 1 'irb3 is more tenacious,
although White's chances are still better.
29 'ird8 +
g7
h6
30 'irg5
White threatened the deadly 3 1 lDf5 + .
31
'irxe5 +
h7
32 h3!
J: d7
White was intending to bring his bishop into play after 33 h2.
But now the denouement comes instantly.
33
lDh5!
.
1-0
The debate about the 12 .*.xf7+ variation still continues.
2 Exchange Variation with

7 Jtc4 - other systems
Game No. 6
Yusupov-Kasparov
USSR Ch 1988
1 d4 6 2 c4 g6 3 c3 d5 4 cd xd5 5 e4 xc3 6 be .i.g7 7
.i.c4 c5 8 e2 c6 9 .i.e3 0-0 10 0-0 .i.g4 11 3 a5
12 .i.d3
12 .i.xf7 + attained wide popularity after the World Champion

ship match in Seville, but capturing on f7 is, of course, not
obligatory - the bishop may retreat instead. It is most securely
placed on d3, although occasionally 1 2 .i.d5 is also seen.
12
13
14
cd
.:t el (18)
cd
.i.e6
The once fashionable Sokolsky attack, 14 d5 .i.xa 1 1 5 "tl'xa1

f6, has now practically fallen into disuse. Black is the exchange
up and can extricate himself without too much difficulty.
18
B
14
15
.i.xa2
"tlra4
This is more precise than 1 5 d5 at once.
42 Exchange Variation with 7 J.c4 - other systems

15
J.e6
J.d7
16 d5
17
'lfb4 (19)
An alternative is 1 7 'lfa3, whereupon, apart from the more
conventional 1 7 . . . e6, Black has 17
bS!?. After 18 :. fd1, the
reply 18
J:.b8 is rather slow, for example : 1 9 'lf b4 a6 20 l0d4

. . .
J:. e8 2 1 J.e2 J.e5 22 J:. a l ! J.c7 23 l0e6! J.d6 24 'lf xd6! fe 25

'lfe5 1 -0; Piskov-Lputian, Belgrade 1 988. In Ilic-Fercec, Kladovo
1 989, Black introduced the important innovation 18
b4!, giving
the pawn back but obtaining adequate counterplay : 19 'lfxb4 :. b8
20 'lf e l l0b3 21 J:. b l a5 22 .tc2, and now 22 . . . a4 is logical,
with a complicated game.
. . .
19
B
In this diagram we have a standard position in contemporary

practice. Black can now choose between 1 7 . . . e6 (the game
continuation) and 1 7 . . . b6. Before going any further, let us
examine the latter.
1 7 . . . b6
Safer than 1 7 . . . b5, for example : 18 : fd 1 J.e5 19 J.c5 b7
20 J.xe7 'lfb6+ 2 1 h1 aS 22 'lfd2 J:. fc8 23 f4 J.g7? (the bishop
should have been placed on d6) 24 e5 b4 25 J.c4 b3 26 J.a3 !
with a won position for White; Balashov-Sibarevic, Lugano 1 988.
18 f4
Or 1 8 .ta6 J.c8 1 9 J.b5 a6 20 J.d3 b5 2 1 f4 e5! 22 J.c5
J:. e8 23 J.d6 'lfb6+ 24 J.c5 (24 h 1 J.g4! 25 J.c7 'ife3) 24
. . . 'lfd8 25 J.d6, with a draw; Savchenko-Lputian, Tallinn 1 988.
18 . . . e5
Stronger than 18
e6 ( 1 8 . . . J:. c8 1 9 l0c3 l0b7 20 J.a6 l:. c7
2 1 e5 l0c5 22 J.c4 'lfb8 23 'lf a3! J:. cc8 24 J:. fd 1 J:. fd8 25 .ixc5!
. . .
Exchange Variation with 7 J.c4 - other systems 43

and Black is in serious trouble; Vyzhmanavin-Ivanchuk, Tashkent
1 987) 19 d6 eS 20 f5 : c8 2 1 .!Llc3 J.c6 22 .!Llb5 1Wd7 23 f6! with
a clear plus; Balashov-Hansen, Malmo 1 987/8. However, after 1 9
.!Llc6 20 1Wb3 e5! 21 f5 : c8 22 .!Llc3 .!Lld4 23 J.xd4 ed 24 .!Lld5
: xc 1 25 .!Lle7 + h8 26 : xc 1 gf 27 ef J.f6, Black has everything
in order; Dolmatov-Gavrikov, Kiev 1 986.
19 :j2
In Vaiser-Gavrikov, Tallinn 1 988, Black seized the initiative
after 19 fS?! :e8 20 J.a6? J.f8 2 1 1fc3 b5. At move 20, White
should have played 20 d6 J.f8 2 1 J.c4 .!Llxc4 22 1fxc4 1Wf6 23 fg
(23 .!Llc3? : ec8 24 gf : xc4! 25 gh + xh7 26 : xf6 J.e6 27 .!Llb5
: xe4, with advantage to Black; Heinig-Gauglitz, Berlin 1 988) 23
. . . 1fxg6 24 : f3 1fe6, with complex play (Gauglitz).
19 . : c8 20 : cf1 1Wc7
It is hard to give a preference to either side.
. . .
17
e6
This position occurrred as far back as 1 954, in the game Geller

Liliental from the 2 1 st USSR Championship. Play continued 18
de J.xe6 1 9 : fd l b6 20 J.a6 1Wh4 21 .!Lld4, and now by means
of 2 1 . . . J.c8 22 J.b5 J.e5 23 g3 1Wf6 24 f4 J.d6 25 1f a4 J.g4
Black could have kept his extra pawn with a sound position.
Nor does 18 d6 give White anything, for example : 1 8 . . . .!Llc6
1 9 1fxb7 : b8 20 1f c7 : b3! 21 : fd 1 .!Lle5 with the initiative;
Razuvayev-Lputian, Sochi 1 987.
In the same Soviet Championship as the game we are annotating
( 1 988), Belyavsky-Kasparov continued as follows : 18 .!Llc3 ed ( 1 8
. . . b6 is also interesting: 1 9 f4 ed 20 .!Llxd5 J.e6 2 1 : fd 1 J.xd5
22 J.b5 1ff6 23 : xd5 : ac8 24 : xc8 : xc8 25 e5 1fe6 26 1We4
J.f8 27 J.d7 : c4 28 1Wd3 1fe7 29 e6, with approximate equality;
Naumkin-Krasenkov, Vilnius 1988) 19 ed (Black similarly has a
good game after 1 9 .!Llxd5 J.e6 20 : fd l J.xd5 etc.) 19
: e8
( 1 9 . . . b6 is also playable : 20 .!Lle4 : e8 2 1 J.d4 .!Llb3 22 J.xg7
.!Llxc l 23 1Wd4 : xe4! 24 fe .!Llxd3 25 J.h6 f6, with a good game;
Utemov-Obodchuk, USSR 1 988) 20 J.f2 J.f8 2 1 1Wb2 (after 2 1
1Wf4 g5! 2 2 Wg3 .!Llb3 2 3 : b1 .!Llc5 24 J.c2 f5! Black would seize
the initiative, so White is not justified in playing for a win here)
2 1 . . . J.g7 22 1W b4 J.f8 23 1Wb2 t-t .
The next two examples are identical with Belyavsky-Kasparov
up to move 20.
Yusupov-Timman, Rotterdam 1 988, continued sharply with 20

44 Exchange Variation with 7 J.. c4 - other systems

J.. e5 2 1 .!tJe4 .tfS 22 .tbS .C. f8 23 J.c5 b6! (not 23 . . . .txe4?
24 fe 11Vh4 2S g3 .txg3 26 .C. c2! and White has an obvious plus)
24 .txf8 11Vxf8 2S 11Vxf8 + xf8 26 .C. fe 1 ! (after 26 .C. fd 1 ? the
exchange on e4 is good for Black : 26 . . . .txe4 27 fe e7) 26 . . .
.!tJb3 27 .C. cd l .!tJd4 28 .ta6 .!tJc2 29 .C. e2 .!tJd4 30 .C. ee l .!tJc2 3 1
.C. e2 .!tJd4 t - t.
One other game by Yusupov likewise ended in a quick draw :
20
b5! 2 1 .C. fd l (or 2 1 .!tJe4 .!tJb7 22 .txbS aS 23 11Va4 .txbS
24 11VxbS .!tJd6, with equality; Naumkin-Mokry, Nemestovo 1 987)
21 . . . .!tJc4 22 J.. xc4 aS 23 11Vb3 be 24 11Vxc4 .C. c8 2S 11Vd3 ! -t;
Yusupov-Smejkal, Munich 1 988.
. .
. . .
18
.C. fd1
It seems to me that a very strong move here is 18 .!tJd4, but it

doesn't appear to have occurred in practice yet.
18
ed
.C. e8
19
ed
b5!
20
.tf2
This looks like a serious weakening, but Black secures for his
knight (via c4) the shortest route to the centre, where the main
action is going to take place.
.!tJc4
21
.!tJd4
.!tJc6
22
By exchanging the troublesome knight on c4, White would
restore the material balance but could hardly count on an
advantage : 22 J.xc4 aS 23 11t'cS be 24 11Vxc4 a4. Yusupov
endeavours to extract the maximum from the position.
22
.txc6
.!tJb2! (20)
23
de
20
w
Exchange Variation with 7 J.c4 - other systems 45

24
J.xbS!
White would lose after the hasty 24 c7? 'irxd3 25 l: xd3 etlxd3
26 cS('ir) : axeS 27 : xeS : xeS and 2S . . . l: c l + .
24
25
c7
etlxdl
'irdS!
It looks as if both sides are playing to win. As Kasparov has

pointed out, a draw would result from 25 . . . 'ireS 26 J.xeS etlxf2!
(26 . . . etlc3? 27 l: e l ! 'irxc7 2S l: e7) 27 'ire7 etlh3 + 2S gh 'irxeS
29 'irxeS + : xeS 30 cS('ir) J.d4+ 3 1 g2 : xeS 32 : xeS + g7.
26
J.xe8
The ingenious 26 J.c6 is answered by the calm 26 . . . 'ire6 27

J.xeS etlxf2, and now 2S 'irbS would lose to 2S . . . 'ire3!. On the
other hand, 26 'irc4 'ir xc4 27 : xc4 etlxf2 2S J.xeS etlh3+ 29 fl
: xe8 30 c8('ir) : xeS 31 : xeS + J.fS 32 gh g7 leads to a draw.
26
27
etlxf2
c8('ir)
27 J.c6 would be met by 27 . . . etlh3 + . Two other variations

also lead to drawn endgames : 27 J.xf7 + xf7 2S 'irf4+ e7
29 cS('ir) : xeS 30 : xeS etlh3+ 3 1 gh 'ir d 1 +, or 27 'irc4 etlh3+
2S fl 'irxc4+ 29 : xc4 : xeS etc., as in the note to White's 26th
move.
27
28
: xc8
: xc8
et:Jh3 + ! (21)
21
w
The knight has sped right the way across the board to get its
king out of trouble.
29
gh
29 fl ?? loses to 29 . . . 'ird3 + 30 e 1 'ire3 + 3 1 dl etlf2+

32 c2 'ird3+ 33 c1 'ir d l mate.
46 Exchange Variation with 7 .t c4 - other systems

1i'dl +
29
30
g2!
And now, 30 f2?? is answered by 30 . . . .td4+ 3 1 g3

1i' g 1 + 32 f4 1i'e3 + 33 g3 1i'gS mate.
30
1i'e2+
gl
31
i-t
There is no escaping the perpetual check: 31 g3 .teS + 32

f4 (32 g4? fS +) 32 . . . 1i'e3 + 33 g2 1We2 + 34 g1 1i'd 1 + 3S
g2 1i'e2 + .
Game No. 7
Polugayevsky-Kudrin
New York 1989

1 d4 f6 2 c4 g6 3 c3 d5 4 cd xd5 5 e4 xc3 6 be .tg7 7
.tc4 c5 8 e2 c6 9 .te3 0-0
10 :t el (22)
In the present game we are still dealing with the main system
(Exchange Variation) of the Griinfeld, but turn our attention to a
line in which White postpones castling by one move, resulting in
play of a wholly different character.
In passing, we should also mention the innovation 10 .:t b 1 !?,
employed in Korchnoi-Kasparov, Reykjavik 1 988. There fol
lowed : 10 . . . aS 1 1 .td3 cd 1 2 cd b6 1 3 0-0 (immediate kingside
activity with 1 3 h4 could come up against counterplay with
obscure consequences : 1 3 . . . eS 14 dS fS 1 S hS f4 etc.) 1 3 . . . e6
14 'it' a4 ( 1 4 'it'd2 .tb7 1 S h4 .:t c8 1 6 .:t fc 1 1i'd7 17 hS : xc 1 + 1 8
.J:l xc 1 .J:l c8 leads to equality) 1 4 . .tb7 1 S .:t fd 1 .l:t c8 1 6 .td2
c6 17 .tc3 1i'h4! 18 .te l .D. fd8 19 f3 1i'e7 20 .tbS?! (in
Kasparov's view 20 .tf2 is better, leading to equality after 20 . . .
aS, but then Black also has 20 . . . .th6!?) 20 . . . a6! 2 1 .txa6
.txa6 22 1i'xa6 xd4 (the initiative is with Black, but after 23
xd4 .txd4+ 24 h 1 , or 24 .:t xd4 1i'cS 2S .tf2, White has
adequate defensive resources; instead, he commits the decisive
error) 23 .tf2? : a8! 24 1i'd3 (24 1i'c4 bS) 24 . . . .:t a3, winning the
queen and with it the game (2S xd4 .:t xd3 etc.).
. .
10
cd
The more placid 10

.td7 is also payable. In Vyzhmanavin
Mikhalchishin, Moscow 1 989, there followed: 11 o-o J:t c8 12 'it' d2
1Wa5 13 d5 e5 14 .tb3 c4?! l S .tc2 e6 1 6 .:t b l ! b6 1 7 f4 g4
. .
Exchange Variation with 7 J.. c4 - other systems 47

22
B
1 8 de fe 1 9 e5! and White gained a clear plus. At move 1 4, Black

introduced a valuable novelty in Shirov-Epishin, Daugavpils
1989: 14
e6! 15 f4 .!Dg4 16 c4 ( 1 6 de J.xe6 17 J.xe6 fe 18 J.f2
: cd8 19 1rc2 .!Dxf2 20 : xf2 1r a6, with the better prospects owing
to the threat of . . . 1rc4) 1 6 . . . 1rxd2 1 7 J.xd2 ed 1 8 cd J.b5 1 9
..tc4 J.xc4 20 : xc4 b 5 2 1 : c2 .!Df6 2 2 .!D g 3 (in Shirov's opinion
White could have maintained equality with 22 .!Dc3 b4 23 e5 be
24 ef cd 25 fg xg7 26 : xd2, or 24 . . . J.xf6 25 J.xc3) 22 . . .
: fe8! 23 d6 .!Dd7! and the initiative passed to Black.
. .
11
cd
1ra5+
J.d7
12
n
An inferior choice is 12
: d8 13 h4 h5 14 1rb3 e6 1 5 d5 .!DeS
. . .
16 de .!Dxc4 17 ef+ h7 18 1rxc4 J.g4 19 f3 : ac8 20 1rb3

: xc 1 + 2 1 J.xc l J.e6 22 1r xe6 :t d 1 + 23 f2 : xh 1 24 1rd6!
'iltxa2 25 1re7!. However, 12
1ra3!? deserves attention. This
novelty occurred in Savchenko-Dimov, Varna 1 989. After 1 3 1rb3
1rxb3 14 J.xb3 :t d8 15 d5 .!Da5 16 J.a4 e6 1 7 . J.g5 f6 18 J.f4
e5 1 9 J.d2 b6, the game is about level.
13
h4 (23)
. . .
23
B
48 Exchange Variation with 7 J. c4 - other systems

13
l:l. ac8
In the diagram position Black has various options. Let us look

at them.
h5 is inadequate; Polugayevsky gives
The blocking move 13
14 f4 h7 1 5 e5!? l:l. ac8 16 J.d3 g8 17 e6! fe 1 8 l:l. c5! with
powerful threats.
Polugayevsky-Korchnoi, Haninge 1 988, went 13 . . . e5 14 d5,
and Black gave up a pawn to no avail : 14 . . . d4 (14 . . . e7 is
safer) 15 xd4 ed 16 J.xd4 J.xd4 17 1hd4 l:l. ac8 18 e2! (but
not 1 8 gl? on account of 18 . . . b5!). Black has no compensation
for the pawn, and the game ended quickly: 18 . . . l:l. fe8 19 f3 l:l. xc4
20 l:l. xc4 11fxa2 + (20 . . . J.b5 doesn't help : 2 1 e3 'irxa2 22 : ee l
'irxg2 2 3 1W d 2 1Wg3 2 4 'irh2, etc.) 2 1 e3 'irxg2 2 2 l:l. hc l f5 2 3 e5!
g5 24 hg 'irh2 25 e6 b5 26 l:l. c7 1 -0.
Polugayevsky won in confident style. The main game we are
examining will also end quickly in his favour. Does this mean that
the outlook as a whole is grim for Black? No - it turns out that
13 . . . l:l. fc8! is a good deal more accurate. After the end of the
main game, we shall consider what a difference it makes to occupy
the c-file with the other rook.

14
h5
e5
Without this central advance Black can hardly hope for counter
play. Let us look at the alternatives :
(a) 14
e6 1 5 hg hg 1 6 e5! (but not immediately 1 6 'ird3 b5 1 7
J.b3 b4 1 8 1W d 2 l:l. xc l + 1 9 xc 1 c6 2 0 'ird3 b4 2 1 e2 l:l. d8,
and Black has very strong counterplay in the centre. Lputian
Dvoiris, Simferopol 1 988, continued 22 J.h6? J.xd4! 23 'irh3
J.f6 24 J.e3 J.c8 25 tt:lf4 !i:Je7, and White got nowhere. 22 g1
was better) 16 . . . e7 17 'ird3 l:l. fe8 18 J.d2 'ira4 19 J.b3 l:l. xcl +
20 J.xcl 'irb4 2 1 'irh3 f8 22 J.h6 g8 23 J.xg7 + xg7 24
'irh8 + f8 25 l:l. h7 e7 26 'lrg7 d8 27 'irxf7, and it is all
over; Griinberg-Gauglitz, E. Germany 1 989.
(b) 14 . . . b5 15 J.b3 e5 (in the present circumstances, this counter
stroke in the centre is not good) 16 hg hg 1 7 de! (Black was vainly
hoping for 17 d5? d4 1 8 l:l. xc8 l:l. xc8 19 xd4 ed 20 J.xd4
J.xd4 2 1 1Wxd4 l:l. c 1 + 22 J.d1 f6 23 g3 'ira4! 24 'irxa4 ba 25
g2 l:l. a l 26 J.f3 l:l. xa2 27 l:l. c 1 a6, when the black bishop settles
on b5 and the pawn on a4 acquires formidable strength) 17 . . .
xeS 1 8 .l: xc8 J.xc8 ( 1 8 . . . l:l. xc8 1 9 'f4 g4 20 J.xf7 + !) 1 9 f4
J.b7? (loses at once, but then Black also has a hard time in other
. . .
Exchange Variation with 7 J.. c4 - other systems 49

vanat10ns : 1 9 . . . l:l. e8 20 xg6 xg6 2 1 "if h5 J.. e 6 22 "ifh7+
f8 23 J.. c 5 + e7 24 J.. x e6 fe 25 l:l. h3, and wins; or 1 9 . . . b4
20 d5 l:l. e8 2 1 J.. g 5 J.. b 7 22 g1 J.. x d5 23 J.. xd5, etc.) 20
xg6! xg6 2 1 "ifh5 1 -0; Dautov-Huzman, Kecskemet 1 989.
Black resigned because he cannot simultaneously defend the knight
on g6 and the h7-square.
hg
15 hg
16 d5!
More energetic than 1 6 J.. d2 1Wb6 1 7 l:l. b 1 "ifc7 1 8 d5 aS! 1 9
J.. d 3 c4, with a good game for Black ; Guseinov-Huzman, Baku
1 988.
16
d4
After 1 7 . . . e7 1 8 J.. g 5, White has a solid advantage.
17
l:l. xc4
17
xd4
eel 18 ..txd4 l:l. xc4 1 9 l:l. xc4 11f a6 ( 1 9 . . . J.. b 5 20 J.. x g7,
and wins) 20 1Wd3 transposes into the game continuation. At move

1 8, it might seem that Black can improve with 18 . . . J.. b5 19
J.. xg7 l:l. xc4 ( 1 9 . . . J.. xc4+ 20 l:l. xc4 !) 20 l:l. xc4 ..txc4 + 21 g1
xg7 22 11f d4 + and after 22 . . . f6 23 "if xc4 "if e 1 + 24 "iffl 11f xe4
the endgame is more pleasant for Black (Huzman and Vaikerman).
However, in place of the check on d4, Polugayevsky unearthed
the quiet move 22 "if cl!! (24) .
,
24
B
In spite of his extra piece, Black is helpless :

(a) 22 . l:l. c8 23 "ifh6+ f6 24 "iff4+ e7 25 1We5+ d7 26
l:l. h7 l:l. f8 27 "ife6+ d8 28 1Wd6+ e8 29 "ilb8 + e7 30 d6 + !
(b) 2 2
11f c5 (22 . . . "ilb4 2 3 a3!) 2 3 1i' b2 + ! f6 24 11f xb7+ l:l. f7 25
l:l. h7 + xh7 26 "if xf7 + h6 27 "if xf6, and Black has no defence.
.
50 Exchange Variation with 7 J.c4 - other systems

(c) 22
f6 l3 .l:l h3! (another quiet move, looking for all the world
like a study; instead, the straightforward 23 'lrh6+ f7 24 'lrh7+
e8 2S 'lrxg6+ d8 26 l:l h7 'lr e 1 + 27 h2 'lrxf2 28 'lrg7 'lrf4+
leads only to a draw). Now Polugayevsky gives two variations :
(c1 ) l3
'lrxa2 24 'lrh6+ f7 2S 'lrh7 + e8 26 'lr xb7 'lra1 +
27 h2 'IreS + 28 g3 'lrd4 29 'lrc8 + f7 30 'lrd7+ g8 3 1 'lrh7
mate.
(c2) 23
J.a6 24 'lrh6+ f7 2S 'lrh7+ e8 26 d6! 'IreS 27
'lrc7 J.bS 28 .l:l h7, and it is all over.
.
. . .
. . .
18
.l:l xc4
'lra6
As we already know, 18 . . ed 1 9 J.xd4 J.bS loses to 20 J.xg7.

ed
19
'lr d3 !
J.b5 (25)
20
J.xd4
.
25
w
One's first impression is that Black has defended himself while

keeping the extra piece. But the unexpected continuation was :
21
22
23
l4
'lrh3!
g1
'lrh7+
.l:l h6!
J.xc4+
f6
f7
1-0
A fine game. But let us go back to move 1 3, where Black could

have played a little differently with 13
.l:l fc8!. This innovation
was used in Nogueiras-Ljubojevic, Barcelona 1 989. Now, after 14
hS e5 1 S hg hg 16 dS d4 1 7 xd4 ed 18 J.xd4 .l:l xc4 ! 19 .l:l xc4
( 1 9 J.xg7 'lra6!) 1 9 . . . 'lra6 20 'lrd3 J.b5, we reach the same
position as in diagram 2S, with the difference that Black's rook is
on a8, not f8. This little nuance reverses the verdict on the position.
Black has an escape square on f8 for his king, and already holds
. . .
Exchange Variation with 7 i.c4 - other systems 51

victory in his hands, seeing that 2 1 1Wh3 (which settled matters in
Polugayevsky-Kudrin) is unplayable; there is no threat of 1Wh7
mate, and Black simply captures the bishop on d4.
I should add that in answer to 14 h5, Ljubojevic refrained from
14 . . . e5, preferring to retreat to the back rank with his knight :
14
ll:ld8. Incidentally, this retreat too was made possible by . . .
l:l. f8-c8 - the rooks are not now disconnected. There followed : 1 5
hg hg 1 6 i.d2? (a most unfortunate move) 1 6 . . . 1Wa4! (Black
seizes the initiative. Mere equality results from 1 6 . . . 1Wb6 1 7 l:l. b 1
i.a4 1 8 1W e 1 1W f6 1 9 l:l. c 1 ll:le6, while in this line 1 8 . . . 1Wc7?
actually loses to 19 l:l. c 1 1Wd7 20 f3 i.xd4 21 1Wh4 1Wd6 22 i.b4!;
Dragomai:etsky-Krasenkov, Moscow 1 989) 17 i.b3 1Wa6 1 8 g1
1Wd3 1 9 l:l. xc8? (it isn't so clear after 1 9 i.h6) 1 9 . . . l:l. xc8 20 e5
i.e6! 2 1 fl i.xb3 22 ab l:l. c2! and Black won.
One month later, the position after 1 3 . . . l:l. fc8 arose in a
game Polugayevsky-Ftacnik, Haninge 1 989. Polugayevsky now
innovated with 14 e5!, and only after 14 . . . ll:ld8 continued with
1 5 h5. Alas, a third beautiful win was not to be: 1 5 . . . i.b5 1 6
i.xb5 1Wxb5 1 7 h g h g 1 8 g 1 l:l. xc l 1 9 ll:lxcl ll:le6 2 0 1Wg4 l:l. c8
2 1 1Wh4? (in Polugayevsky's view, 2 1 1W e4 would have kept some
initiative for White) 2 1 . . . 1Wb1 22 1Wh7+ f8 23 h2 1Wf5 24
ll:le2 l:l. c2 25 l:l. c 1 ? (this allows Black to pick up a pawn and go
into a won ending; it was essential to play 25 l:l. e1 , with drawing
chances) 25 . . . l:l. xc l 26 ll:lxc l ll:lxd4! 27 f4 1We4 28 1Wh3, and now
instead of 28 . . . ll:lf5 (after which Black took 35 moves to win),
Ftacnik demonstrated a quicker method : 28 . . . ll:le6 29 i.d2 g5
30 1Wc3 g8 3 1 g3 gf 32 gf ll:lxf4 33 1Wc8+ h7 34 1Wg4 i.xe5
etc.
So the keen debate about the fascinating variation 10 l:l. c 1 is
far from being resolved.

3 Modern Exchange Variation

Game No. 8
Shirov-Akopian
Tbilisi 1989
lDf6
g6
dS
d4
c4
lDc3
lDf3
cd
e4
be
: bt
l
3
4
5
6
7
s
9
g7
lDxdS
lDxc3
c5
o-o
el
It is interesting that the 1 976 edition of ECO gave this set-up
no more than a cursory mention. Yet in the last few years, the
system with 8 : b 1 has virtually become the most popular choice
against the Griinfeld Defence, perhaps even surpassing the old
variation in which the bishop is brought out to c4.
Another method of developing the white pieces, behind that
strong pawn centre which characterises the Exchange Variation,
involves an early c1-e3. In this case, the light-squared bishop
is brought to e2 (later than usual), and the queen's rook may
occupy either c1 or d1 (after 1Fd1-d2). This variation was an
object of dispute in my last match with Kasparov - see Games
No. 14 and 1 5.
I would add that the popularity of the variation with : b 1 and
.te2 is above all associated with the names of the young
grandmasters "Boris Gelfand and Alexander Halifman.
9
1FaS
The main line for Black is 9 . cd 10 cd, and only then 10 . . .

1Fa5 + . We shall deal with it in Games 1 1 - 1 3. The fairly popular
moves 9 . . b6 and 9 . l0c6 are examined in Games 9 and 10.
. .
Modern Exchange Variation 53

As for 9 . . . .i.g4, it has gone out of use. White gains the advantage,
either with the immediate 10 J:l. xb7, or after sacrificing the d
pawn : 10 0-0 cd 1 1 cd .i.xf3 1 2 ..txf3 .i.xd4 ( 1 2 . . . 11fxd4 1 3
11fxd4 ..txd4 1 4 J:l. xb7) 1 3 J:l. xb7.
10
0-0
Of course, sacrificing the pawn on a2 is part of White's plan.

Whether he will be able to work up a dangerous initiative in
return is another question. Incidentally, at the beginning of the
1 980s White used to prefer 10 J:l. b5 11fxc3 + ( 1 0 . . . 11fxa2 1 1 J:l. xc5
is worse for Black) 1 1 .i.d2 11fa3 1 2 J:l. a5 11fb2 13 J:l. xc5, and at
last Black would take the pawn with 13 . . . 11fxa2. White would
then continue 14 J:l. a5, 14 11f c 1 , or 14 0-0, leading in all cases to
a complex game with chances for both sides.
11f xa2 (26)
10
At this point, capturing the c-pawn is extremely risky. White's
best reply to 10 . . . 11V xc3 is 1 1 d5. Let us see how play might
develop: 1 1 . . . 11fa5 (if 1 1 . . . l'hd7, then 1 2 .i.g5 J:l. e8 1 3 11f a4 is
unpleasant, while 12 . . . .i.f6 loses virtually by force : 1 3 .i.d2
11fa3 14 11fc2 l'he5 1 5 J:l. b3 l'hxf3 + 16 gf 11fa4 1 7 .i.b5 11fd4 1 8
..tc3) 1 2 .i.g5 11fc7 1 3 Wf c l (another good line is 1 3 11fd2 .i.g4
14 J:l. fc 1 b6 1 5 h3 ..txf3 1 6 ..txf3 e5 1 7 d6 11fc6 1 8 -*.e7 J:l. e8 1 9
.i.e2 l'hd7 2 0 .i. b 5 11fb7 2 1 11fc2; Legky-Veingold, Lvov 1 984) 1 3
. . . ..tg4 14 .i.f4 11fc8 1 5 e 5 t2J d 7 1 6 J:l. e 1 (or 1 6 11fe3, connecting
the rooks and increasing the pressure in the centre) 1 6 . . . l'hb6 1 7
d6 ed 1 8 ed, and in return for the pawn White has a highly
promising position; Chiburdanidze-Malaniuk, Odessa 1 982.
26
w
11fe6
11
.i.gS
An interesting alternative is 1 1
t2Jd7 12 .i.xe7 J:l. e8 13 .i.d6
cd 14 l'hxd4 l'hf6 15 f3 l'hd5!? 1 6 J:l. a 1 11f b2 1 7 11fa4 .i.f5 1 8 ed

.i.xd4+ 19 lrxd4 .D. xe2 20 g4 .D. d2 2 1 lrb4 .i.d3 22 lr xb2 .D. xb2
23 .D. fd 1 , with equality; Vaiser-Dvoiris, Bamaul 1 984. Vaiser
suggests 15 .J:l a1 lrb2 16 .lObS!? .ll:l xe4 17 .i.a3 ird2 1 8 .ll:l c 7.
12
ird3!?
A valuable idea, first employed in Epishin-Henkin, Bamaul

1 988. Formerly, White used to advance one of his central pawns
here. For example : 12 eS .D. d8 1 3 lr a4 lrc6 ( 1 3 . . . .i.d7 14 dS
.i.xa4 1 5 de f6 1 6 ef ef 1 7 .i.c4! .D. e8 1 8 .i.f4 b6 19 .i.dS .i.c6
20 c4! gave White a clear advantage in Yusupov-Tukmakov,
Moscow 1 983) 14 irb3 .i.e6 ( 1 4 . . . lrc7 1 5 .i.c4 .D. f8 1 6 e6 f6 1 7
.i.h4 .ll:l c6 1 8 .i.g3 ird8 1 9 lr a2 cd 20 cd b 6 2 1 d S .ll:l e S 2 2 .D. fd 1
h8 2 3 .ll:l x eS fe 2 4 d6! gave White a large plus in another game
Yusupov-Tukmakov, Erevan 1 982) 1 5 c4 cd 1 6 .i.xe7 .D. c8 1 7
.ll:l xd4 .i.xc4! 1 8 .ll:l x c6 .i.xb3 1 9 .ll:l x b8 .i.e6 20 .i.d6 .i.f8 2 1
.i.f3 .i.xd6 2 2 .i.xb7 .i.xb8 2 3 .i.xa8 .i.xeS 24 .i.b7 .J:l c2 2 5
.D. fc 1 .J:l a2 2 6 .l:t d 1 .i.c7 2 7 .J:l a 1 .D. b2 2 8 .l:t db 1 .l:t d2 2 9 .l:t d 1 .l:t b2
1-1 ; Damjanovic-Schmidt, Athens 1 9 84. This one game serves
to illustrate the harmlessness of 1 2 eS.
12
b6
After 12
lrd6 1 3 lre3, or 12
.ll:l d7 1 3 lre3 .ll:l f6 14 .ll:l e S!,
the initiative is with White. On the other hand, 12
.D. d8!?
deserves to be tried.
. .
. .
13
d5
In this case, 1 3 lre3 is no good : 1 3 . . . .i.a6! 14 dS 1rd6 1 5 c4

e5!, and White has nothing for the pawn.
13
ird6 (27)
27
w
14
e5!
The point of White's play. He now obtains the two bishops and

a big advantage in space and development. Still, Black does have
two extra pawns.
Again 14 1We3 would be well answered by 14 . . . eS!.
14
IS
16
et:lxeS
1W d 2
-*.xeS
'itxeS
But not 1 6 f4? 1Wd6 1 7 1We3 : es, or 1 6 1We3? 1Wxe3 1 7 fe : eS!

1 S d6 eS! 1 9 -*.e7 -*.e6 20 .i.f3 ll!d7 21 -*.xaS : xaS.
16
1W d6
Perhaps 1 6 . . . et:ld7 does more to limit White's options. The
virtually forced continuation is 1 7 -*.f3 1Wd6 1 S : fe 1 f6 (worse
alternatives are 1 S . . . et:leS? 19 : xeS! 'itxeS 20 : e 1 W'fS 21 d6
-*.d7 22 de : res 23 -*.xaS : xaS 24 h3, and 19 . . . f6 20 l:t e6!
-*.xe6 2 1 -*.f4 1Wd7 22 de 1Wxd2 23 -*.xd2) 1 9 : e6 fg 20 : xd6
ed 21 l:t e l aS, and the position is unclear.
17
1We3!
Here this move is entirely in place, and better than 17 .i.f3

ll!d7 (see the previous note), or 17 -*.f4 eS 1S de 1Wxd2 19 -*.xd2
fe 20 .tf3 : xf3 ! 21 gf .tb7.
17
: e8
But not 1 7 . . . f6 1 S .tf4.

18
.tf3
lll d 7
The play hinges on some subtle points. In many lines the
counter-stroke . . . e7-eS is good, but not here : 1 S . . . eS? 19 de
: xe6 20 : bd 1 ! and wins.
19
.tf4
It is only now that White deviates from the game Epishin
Henkin already referred to. That game went 19 : ret et:lf6 ( 1 9 . . .
f6 20 1We6 + !) 20 c4 .tfS 21 : at, and after 21 . . . a6?! 22 -*.f4
1Wd7 23 h3 hS? 24 .teS h7 2S 1Wf4 l:t gS 26 'itgS, White quickly
won. But instead of 21 . . . a6 Black has 21
eS!, after which
White doesn't have adequate compensation for the two pawns,
for example: 22 de : xe6 23 .txaS (43 1W c3 et:le4!) 23 . . . : xe3 24
fe. In Epishin's view, White can improve with 21 l:t bd l -*.g4!
(otherwise 22 h3) 22 .txf6 .txf3 23 .teS .txd l 24 .txd6 ed 2S
1Wc3, but Henkin has indicated a line in which Black again has noth
ing to fear: 25 . . . .te2 26 f3 bS! 27 cb .txbS 2S : xeS + : xeS, 29
1Wf6 : e 1 + 30 f2 : e2 + 3 1 g3 c4 32 1Wxd6 a6. But then, the
text move should not have brought White any concrete gains
either.


19
20
21
Wf6
: b8
d6
: bd1 (28)
2 1 de is bad : 2 1 . . . :Z. xe7 22 Wd2 .!LieS 23 ..tgS .!Llxf3 + 24 gf

:Z. d7! 25 Wxd7 WxgS + , and Black wins.
28
B
eS?
21
A serious mistake. He could not have solved his problems with

21
..t b7 22 de :Z. xe7 (22 . . . ..txf3 23 :Z. xd7 ..tg4 24 : xa7
:Z. bc8 25 :Z. e l ) 23 :Z. xd7! : xe3 24 fe ..txf3 25 ..txb8 W xc3 26
:Z. xf3 Wel + 27 : fl Wxe3 + 28 ..t h l We8 29 :Z. fd l ! and wins. But
the essential move was 21
e6!, and already after 22 ..tc6 Wd8
23 :Z. fe l bS it is White who has to think about equalising.
W g7
22
..tg5
If 22 . . . Wf5, then 23 ..te4 We6 25 ..tdS WfS 25 f4!, while 22
. . . Wh8 is well answered by 23 ..tc6.
. .
. .
23
24
..th6
..tc6!
Wf6
:Z. e6?
24 . . . Wd8 25 f4 ..tb7 26 ..txd7 Wxd7 27 fe f5! was more

tenacious, although after 28 ..tgS ..te4 29 ..tf6, White has an
advantage due to the subsequent h4-h5.
25
26
..tg5
f4
W g7
h6
If 26 . . . f6, then 27 fS! gf 28 ..th6 W g6 29 :Z. f3 f4 30 :Z. xf4!

27
28
29
30
31
32
..te7
W xf4
Wa4
..td5
W xa7
..txb7
1-0
ef
g5
.!Oe5
..t b7
.!Lld7

Game No. 9
Halifman-Pribyl
Leningrad 1989
1 d4 ltlf6 2 c4 g6 3 ltlc3 d5 4 ltlf3 J.g7 5 cd llJxd5 6 e4 ltlxc3
7 be c5 8 l:l. b1 0-0
9
J.e2
b6
10
0-0
Another familiar plan is 10 J.e3 J.b7 1 1 e5 llJc6 1 2 h4,
with a kingside attack. However, in a game Nemet-Korchnoi,

Switzerland 1 985, Black seized the initiative with ll
cd! 12 cd
ltla6 1 3 1rd2 ltlc7 14 h4 ltld5 1 5 h5 llJ xe3 1 6 fe e6 1 7 .i.d3 l:l. c8.
There is also another possibility (after 1 1 . . . cd 12 cd): 1 2
J.e4!
13 l:l. c l llJc6 14 h4 l:l. c8 1 5 llJg5 J.xg2 1 6 l:l. h2 J.d5 1 7 h5 llJb4
1 8 hg hg 1 9 a3 llJa2, and Black is clearly better; Vaiser-1. Sokolov,
San Bernadino 1 989.
In Cebalo-1. Sokolov from the same tournament, White
launched his attack at once with 10 h4!?. His opponent's ineffective
response unexpectedly led to a quick and crushing defeat : 1 0 . . .
J.g4 ( 1 0 . . . J.b7 is more logical) 1 1 '/o>fl !? cd 1 2 cd llJd7 (and
here, the right continuation is 1 2 . . . llJ c6 1 3 d5 J.xf3 14 gf ltld4)
1 3 e5! l:l. c8 14 J.e3 llJb8 1 5 h5! 1rd5 1 6 hg hg 1 7 1rd2 J:Hd8 1 8
l:l. h4 ! .i.xf3 1 9 gf J.xe5? 20 l:l. b5! 1re6 2 1 l:l. xe5, and i t i s all over.
. . .
10
11
J.b7
1rd3 (29)
29
B
11
J.a6
In earlier games, Black played 1 1 . . . cd 12 cd e6 13 .i.g5 1rd6,

but as a rule failed to equalise. Here is one example : 14 'iWe3 l:l. c8
1 5 l:l. fd 1 llJd7 1 6 h4 l:l. c2 1 7 e5 1rc7 1 8 J.d3 l:l. c3 ( 1 8 . . . l:.Xa2 1 9

l:t bc 1 ) 19 'lre2 .*.d5 20 : be l h6 2 1 .*.e3 l:t dS 22 h5! with a big
advantage; Lputian-Lalic, Sarajevo 1 9S5.
12
'lre3
1 2 'lrc2 promises White less. Chernin-Dorfman, Moscow 19S4,

continued 1 2 . . . cd 1 3 cd 'Ir eS 14 'lr d l .*.xe2 1 5 'lrxe2 'lra6 1 6
'lr xa6 lbxa6 1 7 .*. a 3 l:t fdS 1 S d5 f5 19 .*.xe7 l:t d7 2 0 d6, and the
players agreed a draw.
12
. .
12
cd
e6 is too timid; after 13 de .*.xe2 14 'lrxe2 be 1 5 .*.f4
'IreS 1 6 .*.d6 :t dS 1 7 e5 'lrc6 l S h4 lbd7 1 9 c4! a6 20 : fe l lbb6

21 'lre4, Black has no prospects whatsoever; Lerner-Gorelov,
Ivano-Frankovsk 1 9S2.
In the last few years, Black has tried moving his queen to cS or
d7 without exchanging pawns in the centre. These experiments
have been unsuccessful, for example : 12
'IreS 1 3 d5 .*.xe2 1 4
'lr xe2 .*.xc3 1 5 e5 'lrf5 1 6 : b 3 .*.a5 1 7 lbh4 'lr d 7 1 S .*.h6, with
advantage; Halifman-Lau, Rotterdam 1 9SS. Or 12
'lrd7 1 3 de
be ( 1 3 . . . .*.xe2 14 'lrxe2 be is safer) 14 l:t xbS!? l:t axbS 1 5 .*.xa6
'lr a4 16 'lre2, with a won position; Gelfand-I. Sokolov, Yugoslavia
1 9SS.
'lrd7
13
cd
A recommendation of Grandmaster Gavrikov. A much earlier
game Browne-Martz, USA 19S2, went 1 3 . . . .*.xe2 14 'lrxe2 llJc6
1 5 d5 lbd4 16 llJxd4 .*.xd4 1 7 l:t d l .*.g7 1S .*.a3 'lrd7 19 e5,
with a clear advantage to White.

. . .
14
.*.a3
14 .*.xa6 lbxa6 1 5 'lra3 lbc7! gives White nothing; Halifman
Epishin, Leningrad 1 9SS, continued 1 6 :t d l :t fcS 1 7 .*.b2 lbb5

lS 'lrd3, and now Halifman recommends lS . . . e6!
Black also obtains a good game after 14 l:t dl?! 'lra4 1 5 .*.d2
.*.xe2 16 'lrxe2 :t c8 17 e5 lba6 18 e6 f6 ( 1 8 . . . f5!) 19 d5 'ifc4 20
'lrxc4 : xc4; Lputian-Zilberstein, Blagoveshchensk 1 988. How
ever, Salov's innovation 14 d5!? deserves attention. After 14 . . .
.*.xe2 1 5 'lrxe2 'lra4 1 6 .*.g5 l:t e8 1 7 l:t fc l , White obtained a
slight edge in Salov-1. Sokolov, Haifa 1 989.
.*.xel
14
Another possibility is 14
l:t e8 IS d5 .*.xe2 16 'lrxe2 'lra4 1 7
'lre3 lbd7 l S l:t fc 1 : ac8 1 9 h 3 l:t xc l + 2 0 l:t xc l llJf6 2 1 lbd2
.*.h6!, as in Gelfand-Malishauskas, Vilnius 1 988. However, Pribyl
gives IS : fd! as more accurate.


15
'ir xe2
.l:l. e8
Black is not in a hurry to bring his knight out. In Gelfand

Groszpeter, Palma de Mallorca 1989, Black played 1 S . . . c!Llc6,
and after 16 dS c!LleS 17 c!LlxeS -*.xeS 1 8 f4 d6 19 .i. b2 e6 20
.1:1. bd 1 White gained the initiative and soon won.
16
17
.l:l. fcl
.l:l. c3
e6
.l:l. c8
1 7 . . . c!Llc6? loses to 1 8 'irbS .l:l. ac8 1 9

18
.I:!. be l
a6 20 'ir xb6.
.l:l. xc8+
1 8 .l:l. bc 1 .l:l. xc3 1 9 .l:l. xc3 c!Llc6 20 "irbS .l:l. c8 2 1 .l:l. c4 'irb7 leads
to equality.
18
'ir xc8
19
.l:l. c l
'irb7
An alternative is 19 . . . 'ira6 20 "ire3 c!Lld7 2 1 .l:l. c7 c!Llf6 (2 1 . . .
jfa4 22 h3) 22 c!LleS .l:l. c8! 23 .l:l. xf7 "irbS 24 g3 jfbl + 2S g2

jfxe4+ 26 jf xe4 c!Llxe4 27 .l:l. xa7 xeS 28 de .l:l. c2, with full
equality; Dzhandzhgava-Malishauskas, Uzhgorod 1 988. A better
line for White, according to Halifman, is 20 jf xa6 c!Llxa6 2 1 fl.
with some advantage.
20
'irc4
c!Lla6 (30)
30
w
This position had first occurred only a few days earlier, in the
game S. Ivanov-Pribyl from the same international tournament
(Leningrad 1989). After 2 1 dS ed 22 ed .l:l. d8 23 d6 c!LlcS! 24 xeS
be 2S 'ir xcS 'ira6 26 h4 .l:l. xd6 27 c!LlgS h6, the players agreed a
draw. The main game we are examining was to end in victory for
Black, but that was hardly the rightful outcome. For the moment
Halifman doesn't want to settle for a draw, and adds fuel to the
flames.

21
22
23
24
25
26
27
28
h4
e5!?
h5
.l:l. c3
'irc6
.l:l. xc6
.l:l. c7
..txf8
29
d5!
.l:l. d8
h6
gh
b5!
'irxc6
lt! b8
.U8
xf8 (31)
31
w
White's last chance - he hopes for success by sacrificing another

pawn. As Pribyl has pointed out, a forced draw results from 29
.l:l. xa7 lt!c6 30 .l:l. b7 lt!xd4 3 1 lt!xd4 .l:l. xd4 32 .l:l. xb5 .l:l. d l + 33 h2
.l:l. d2.
29
30
31
32
33
lt!d4
f4
.l:l. c5
.l:l. xd5
ed
.l:l. e8!
.l:l. e7
lt!a6
In the event of 33 .l:l. c6 lt!c7 34 .l:l. xh6, Black can save himself
with 34 . . . g7! (but not 34 . . . lt!e6? 35 lt!f5 .l:l. c7 36 .l:l. h8 mate)
35 .l:l. xh5 lt!e6 36 lt!f5 + g6 37 lt!xe7 + xh5 38 lt!xd5 g4.
33
34
35
36
37
.l:l. d6
lt! c6
f5
f6?
lt!c7
lt!e6
.l:l. c7
lt!g5
Now all White's previous work comes to nothing. After 37 e6!

Black would still have to struggle for the draw.

37 . . . g8 38 fl 'it>h7 39 lll b4 a5 40 lll d5 .:t e l + 41 'it>e2
.:t el + 42 'it>e3 .:t xg2 43 .:t d7 .:t g3 + 44 'it>e2 .:t g4 45 'it>d3 h4 46
'it>e3 .: a4 0-1
Game No. 1 0
Halifman-Henkin
Leningrad 1989
1 d4 lll f6 2 c4 g6 3 lll f3 J.g7 4 lll c3 d5 5 cd lll xd5 6 e4 lll xc3
7 be c5 8 .:t b1 0-0
9 J.e2
lll c6
10
d5
White is unsuccessful with 10 J.e3 cd 1 1 cd 'ifa5+ 1 2 J.d2

'ifxa2; Paunovic-Ristic, Vrnjacka Banja 1 983. In comparison with
the variation 9 . . . cd 10 cd 'ifa5 + , which we examine in Games
1 1 - 1 3, Black has the useful extra move . . . lll c6.
10
lll e5
Safer than taking the pawn. Let us look at that possibility : 1 0

. . . J.xc3 + 1 1 J. d 2 J.xd2+ 1 2 'if xd2 lll a 5 (moving that knight
to b8 or d4 is worse. An example of the latter is 12 . . . ll:ld4 1 3
lll x d4 cd 14 'ii' xd4 'ifa5 + 1 5 'ifd2 'ifxd2+ 1 6 'it> xd2 .:t d8 1 7 'it>e3
b6 18 .:t bc l e6 19 J.c4 e5 20 J.b3 J.d7 21 .:t c7 a5 22 d6, and
Black has numerous problems; Kasparov-Natsis, Malta 01 1 980)
1 3 h4 (White may also castle, with e4-e5 to follow) 1 3 . . . J.g4
14 h5 J.xf3 1 5 gf! (not allowing the knight onto c4). Despite the
simplification, White's attacking potential is very noticeable, for
example : 1 5 . . . e6 1 6 'if h4 'iff6 1 7 hg 'if xg6 1 8 'ifh2 'it>h8 1 9 'it>d2!
f6 20 .:t bg l 'iff7 21 de 'ife7 22 .:t g6, and the attack continues;
Bjarnason-Griinberg, Biel 1 985.
J.xe5
lll xe5
11Vd2 (32)
White can also play 12 c4 at once, for example : 12
11
12
'if d6!?
( 1 2 . . . 'ifa5 + occurs more often) 12 'ifd2 J.d4 14 J.b2 J.xb2 1 5

: xb2 e 5 1 6 0-0 b6 1 7 : b3, with a complicated position; Gelfand
Ghinda, Halle 1 987. Alternatively Black has an interesting counter
sacrifice of a pawn : 12
f5 1 3 ef J.xf5 14 .:t xb7 'if d6 1 5 .:t b3.
Perhaps he should now go in for immediate exchanges with 15 . . .
J.e4 1 6 J.f3 J.xf3 1 7 : xf3 : xf3 1 8 'ifxf3 : f8, trying to exploit
his advantage in development. Danner-Shvidler, Biel 1 982, saw
instead 1 5 . . . J.d7 16 'ifd3 .i.f5 17 'ifd2 J.e4 1 8 J.g4 J.xg2 1 9
J.e6 + 'ifxe6 ( 1 9 . . . h8 20 .:t g 1 J.e4 2 1 .:t h3 with an attack!) 20 de


.i.xh 1 2 1 11fe2 l:tf5 22 f3 .J:I. af8 23 .i.h6 .J:1. 8f6 24 11fe4, and Black
has worries. A move that looks fairly logical is 12
11fc7,
covering the pawn on b7 and preventing White from castling. A
correspondence game Danner-Nesis ( 1 983-5) continued 1 3 h3 f5
1 4 ef .i.xf5 1 5 .J:I. b3 ..td4 1 6 0-0 e5, and Black achieved a sound
position.

32
B
In the first half of the 1980s, Black almost invariably played 1 2

. . . e 6 here, and after 1 3 f4 h e retreated his bishop t o g 7 o r h8. I
do not plan to give a thorough study of the positions which then
arise, but it is worth reminding the reader of the difference that
the choice of retreat square makes - or more exactly, of why it is
essential to withdraw the bishop into the corner.
Consider the variation 12
e6 13 f4 .i.g7 14 c4 .J:I. e8 1 5 e5 f6
1 4 d6 fe 1 7 .i.b2 ef 1 8 .i.xg7 <t;xg7 1 9 0-0 (33) .
. . .
33
B
Practice has abundantly shown that 1 9

.J:I. f8 20 .J:I. xf4 .J:I. xf4
2 1 11fxf4 secures White a distinct plus. But at move 1 9 Black
. . .

cannot play 1 9 . . . eS, on account of 20 d7!, and the pin on the
7th rank will be lethal.
However, if earlier Black plays 13
.th8, and accordingly 1 8
.txh8 xh8 occurs, then 1 9 0-0 can be answered effectively by
19 . . . eS!, giving Black a solid position after 20 :t fd l 'it'd7 2 1 'it'dS
:t b8 22 'it' xcS b6; Bikhovsky-Polovodin, Irkutsk 1983.
All the same, I should point out that after 13 . . . .th8 14 c4
:t e8 15 e5 j6, White doesn't have to play 1 6 d6; he has the
stronger 16 f5! (34) .
. . .
34
B
16 . . . ed
Other pawn captures are no better: 1 6 . . . ef 1 7 e6! followed by
1 8 0-0 and 19 g4; 1 6 . . . fe 1 7 fg ( 1 7 fe 'it' h4+ 1 8 d 1 doesn't
look bad either) 17 . . . hg 18 0-0 ed 19 cd 'it'd6 20 :t b3 .tg7 2 1
:t g3 etc.; 1 6 . . . gf 1 7 :t b3 :t e7, and now 1 8 .tb2, 1 8 d 6 and 1 8
'it'h6 are all highly unpleasant for Black.
1 7 fg
Practice has also seen 1 7 e6 d4 1 8 g4 b6 1 9 .tf3 : b8 ( 1 9 . . .
.ta6!?) 20 'it' g2 'flc7 2 1 0-0 gS 22 .tdS 'it'e7 23 h4, with the
initiative; Vaiser-Pribyl, Sochi 1984.
17 . . . hg, and now White has two lines :
(a) 18 0--0 : xeS 1 9 .tf3 d4 20 :t e l 'it'e8 2 1 .ta3 .tfS 22 : xeS
'it' xeS 23 .tdS + 'it' xdS 24 cd J.. x b l 2S J.. x cS, and Black is in a
bad way; Cebalo-Raicevic, Citorle 1 984.
(b) 18 cd : xeS 19 0-0 b6 20 d6 J..e 6 21 .tf3 :t c8 22 J.. b2 fS 23
'it'h6! and White has a won position; Agzamov-Pribyl, Sochi 1984.
ll
b6
A comparatively new idea. Black limits the effect of his

opponent's rook on the b-file, and makes b7 available to his own
bishop.
13
f4
i.g7
14
. . .
e6
14
c4
e5 looks logical, but the following two examples from
recent practice sho w that it is not so simple for Black to achieve

equality.
Epishin-Ftacni, Belgrade 1 988, went 1 5 0-0 ef 1 6 11Vxf4 i.d4+
17 i.e3 i.xe3+ 18 11V xe3 11Ve7 1 9 11Vf4 (Epishin gives 1 9 e5! l% e8
20 e6 fe 2 1 i.f3 as even stronger) 1 9 . . . f6 20 a4! with the better
game for White.
Gelfand-Fta<::n ik, Debrecen 1 989, deviated with 15
f5!? 1 6
i.b2 1r d 6 1 7 1r c 3 l% e8 1 8 i. d 3 l% e7? (in Ftacnik's view, 1 8 . . .
fe 1 9 i.xe4 l% e7 20 f5 gf 2 1 i.xf5 e4 leads to unclear play) 1 9 ef
gf ( 1 9 . . . e4 20 f6 ed 2 1 fe! and wins) 20 fe! i.xe5 2 1 11Vd2 i.xh2+
22 hl i.e5 23 s5 + 1rg6 24 11Vxe7 1rh6+ 25 gl 11Ve3 + 26
h l 11Vh6+ 27 g l 1re3 + 28 l% f2! i.h2 + 29 xh2 11Vxe7 30
l% f3! 11Vd6+ 31 l% g3 + f7 32 : fl h5 33 i.xf5 h4 34 i.g6+ +
g8 35 i.h7 + + ! xh7 36 l% f7 + h6 37 i.cl + 1-0.
. . .
15
i.b2
Considerably stronger than 15 o-o, which Black used to meet

with 1 5 . . . l% e8. But 15 . i.d4+! is better; after 16 h1 l% e8
1 7 i.d3 ed 1 8 ed i.f5! 1 9 i.xf5 gf, Black obtained good play in
Novikov-Krasenkov, Odessa 1 989. The game continued very
sharply and ended in a draw; here is how : 20 i.b2 1rf6 2 1 i.xd4
cd 22 : be l l% e4 23 g4 l% ae8 24 gf 1rxf5 25 l% g 1 + f8 26 1rd3!
11Vxf4 27 11Va3 + l% 8e7 28 : eft 11Vh4 ! 29 1rd6 l% g4 30 11Vd8 + l% e8

1 3 1rd6+ : e7 32 11Vd8 + l% e8 33 : xf7+ xf7 34 l% fl + l% f4! 35
1rc7 + g8 36 11Vxf4 11Vxf4 37 : xf4 d3! 38 l% d4 f8 39
l% xd3 l% e 1 + t -t .
. .
15
16
17
l% xb2
e5
i.xb2
l% e8
On 1 7 0-0, Black would equalise with 1 7 . . . ed.

i.b7
l7
18
19
20
0-0
i. f3
l% e 1
ed
f6
fe (35)
This position arose for the first time in Kudzhma-Henkin,

Kramatorsk 1989. In that game Black obtained a fine position

after 2 1 .I:!. xeS 11Ff6!. There followed 22 11F e l (22 .*.xd5 + was more
to the point, giving equality after 22 . . . .*.xd5 23 11Fxd5 + h8)
22 . . . .l:l. xe5! 23 11Fxe5 (23 fe 11Ff4) 23 . . . 11rf7 24 .l:l. e2 (better is 24
cd .l:l. e8, when Black has a minimal edge) 24 . . . de 25 .*.g4 11Fg7!
and White's position is not easy.
35
w
21
fe!
A significant refinement.
21
22
23
24
25
.*.xd5+
'it'xd5+
e6
11Ff3
11Fc7
.*.xd5
g7
.l:l. ad8
.l:l. e7
He should have blockaded the pawn with his queen, although

after 25 . . . 11Fe7 26 .1:1. be2 White still has the more pleasant position.
26
27
28
.1:1. 2
11Ff6 +
11Fg5!
11Fd6
g8
This accurate move casts doubt on Black's set-up. After the

immediate 28 h4 .l:l. de8, White would have to acquiesce in a draw
with 29 11Ff7+ h8 30 11Ff6 + . But now 28 . . . .l:l. de8 is answered
by 29 11Fd5! .l:l. d8 30 .l:l. d2! with a won position.
g7
18
29
30
31
32
33
'it'f6 +
11Fg5
h4!
11Fd5
g8
g7
.l:l. de8
.l:l. d8
11Fg5
Of course not now 33 .l:l. d2? 11Fg3!, and it is Black who wins.

33
34
:t dl
:t de8
11fc6?
As Henkin has shown, the only defence was 34

11fc7. Then
35 h5?! :t xe6 36 h6+ f8 37 :t xe6 :t xe6 38 :t d8 + f7 39 :t h8
0 0 0
11fd6 40 :t xh7 + g8 4 1 :t g7+ h8 42 :t xa7 :t e l + ! 43 f2

:t e5 44 :t a8 + would lead to a draw. Still, after the quiet 35 :t del
White would retain some positional advantage.
g8 (36)
35
11fe5+
36
w
36
:t d7!
Black's pieces are stalemated.

36
0 0 0
36
b5
h5 doesn't help : 37 :t e3, followed by 38 :t g3. If 36
0 0 0
11fa4, White's simplest course would be 3 7 h5 11f xc4 38 :t xe7 :t xe7

39 11f b8 + g7 40 h6 + ! xh6 42 11ff8 + :t g7 43 e7.
37
38
39
40
41
42
43
cb
hS
:t xe7
h6
h1
:t b1 !
11f f4!
11f xb5
11fb6
:t xe7
c4+
11fd8
:t c7
After 43 11fd5 :t c8!, or 43 11fd4 'tiffS!, Black could still hold on.
g5
43
The threat was 44 e7!

44
11fxg5 + !
1-0

Game No. 1 1
Vaiser-Pein
Budapest 1989
l d4 lt!f6 2 c4 g6 3 lt!c3 dS 4 lt!f3 ,j_g7 S cd lt!xdS 6 e4 lt! xc3
7 be cS 8 l:t bl 0-0
9
10
,j_el
cd
cd
'lfaS+
At the present time, the exchange on d4 followed by the check

with the queen is Black's most popular choice in this variation.
ll
,j_dl
Formerly, 1 1 'lf d2 was seen more often, but in the resulting

endgame White can hardly count on a plus. For that reason, the
search for the initiative now proceeds on different lines, and
involves a pawn sacrifice. This variation of the Griinfeld occurred
only once in my encounters with Kasparov - in the 1 3th game in
Seville, where 1 1 'lt'd2 'lfxd2 + 12 ,j_xd2 was played (37) . Let us
briefly examine that game.
37
B
1 2 . . . e6 (in answer to 1 2 . . . b6 at once, White has various

choices : 13 ,j_d3, 1 3 d5 or 1 3 l:t c 1 , but best of all is 13 0-0. Then
after 1 3 . . . ib7 14 d5 ,j_a6 1 5 l:t fe l ,j_xe2 16 l:t xe2 lt!a6 1 7 .i.e3,
White has the better chances. If 17 . . . f5, Gutman recommends 1 8
d4! fe 19 l:t xe4 if6 20 lt!e5!. I f 1 7 . l:t fe8, he recommends
18 l:t c2!, instead of 1 8 lt!d4 xd4 19 .i.xd4 e6 20 de, which led
to a draw in Halifman-Tseshkovsky, Minsk 1 985) 1 3 0-0 ( 1 3
.i.c4, 1 3 l:t c l and even 1 3 h4 have been seen, but there i s n o point
in starting active operations with incomplete development) 1 3 . . .
b6 1 4 l:t fd l (it is hard to decide how to arrange the rooks on the
first rank; 14 l:t fe l , 14 l:l. bc l and 14 l:l. fc l have also been played.
.

As the d-pawn is due to advance, I place a rook behind it) 14 . . .
..tb7 1 5 d5 ed 1 6 ed lDd7 ( 1 6 . . . ..txd5 loses to 1 7 ..tb4 .l:l. d8 1 8
..tc4) 1 7 ..tb4 .l:l. fc8 1 8 ..te7 (neither 1 8 tDd4 nor 1 8 ..tb5 gives
White anything. But now his initiative looks fairly substantial for example 1 8 . . . ..tf8 1 9 d6 ..txe7 20 de tDf6 2 1 .l:l. e 1 !. However,
Black manages to find a powerful retort which solves all his
opening problems) 1 8 . . . ..tf6! 1 9 d6 g7! 20 .l:l. e 1 (20 ..txf6+
is answered by 20 . . . xf6, but not 20 . . . lDxf6 2 1 1 d7 .l:l. d8 21
lDe5!) 20 . . . .l:l. c5 2 1 ..tb5 ..tc6 22 ..txc6 .l:l. xc6 23 .l:l. bd 1 ..tc3 24
.l:l. e3 f6 25 g4 g5 26 h4 h6 27 hg hg 28 lDd4 (now Black seizes the
initiative; the correct line was 28 .l:l. c 1 .l:l. ac8 29 ..td8! with equality)
28 . . . ..txd4 29 .l:l. xd4 .l:l. h8 30 .l:l. e l .l:l. c2 (Black could have
continued quite actively with 30 . . . .l:l. c3 3 1 g2 f7; but now
peace ensues quickly) 3 1 a4 a5 32 f4! g6 33 fg xg5 34 .l:l. fl
g6 35 .l:l. f2 .l:l. hc8 36 .l:l. df4 .l:l. xf2 t-t.
11
"Wxal
b6 (38)
12
0-0
The main variations are 1 2 . . . tDd7 and 1 2 . . . "We6; these will
feature in Games 12 and 1 3. There is no point in exposing the
knight to a pawn attack : after 12 . . . lDc6 1 3 d5 lDe5 14 lDd4,
White has clearly the better chances. But then, in the last few
years, practice has shown that 1 2 . . . b6 is also dangerous for
Black. The present game will convince us of this.
38
w
13
"Wet!
"We6
The alternative is 1 3 . . . ..tb7 1 4 ..tc4 "W a4 1 5 ..tb5 "Wa2 (now

of course White can force a draw by repetition, as occurred for
instance in Chemin-Dvoiris, Lvov 1990; but that is not much of
an achievement) 1 6 .l:l. e l ! (but not 1 6 "W e t ? "Wc2! 1 7 d5 ..ta6! 1 8

We3 .txb5 1 9 J: xb5 a6 20 l:l. bb1 c5 2 1 .tb4 aS!, and Black
seizes the initiative; Yusupov-Korchnoi, Reykjavik 1 988) 1 6 . . .
: c8 1 7 W d 1 e6 1 8 We2 c6 1 9 We3 l:l. d8 ( 1 9 . . . e5 is no better :
20 d5 d4 2 1 xd4 ed 22 Wf4 .tf8 23 l:l. a l Wc2 26 .i.d7! with
more than enough for the pawn; Halifman-Epishin, Vilnius 1 988)
20 : a 1 Wb2 21 .txc6 .txc6 22 .tc3 W b3 23 J:l. eb 1 W c2 24 d2
.th6 25 W xh6 W xc3 26 f3 f6 27 l:l. c 1 'lfb2 28 'Wf4! 'it;>g7 29
'Wc7 + .td7, and now instead of 30 'Wd6, as played in Epishin Mark Tseitlin, Leningrad 1 988, Epishin gives 30 : cb 1 ! 'We2 3 1
J:l. e 1 'Wb5 3 2 : xa7 : xa7 3 3 'Wxd8 .te8 3 4 e 5 J: f7 35 h3, with
the advantage.
14
15
16
17
.tc4!
:e1
.tb4
'W xe4
'Wb7
.te6
J: xe6!
A typical exchange sacrifice giving White a dangerous initiative.
17
18
fe
g5 (39)
39
B
18
c6
A novelty which, however, does not solve all Black's problems.

In Gelfand-Dorfman, Minsk 1 986, Black played 18
h8,
which gave him a won position after 19 xe6 d7 20 .txe7?
: fc8. In Vaiser-Andrianov, Naberezhnie Chelny 1 988, White's
play was strengthened : 19 J: b3! d7 20 l:l. h3 f6? (loses at once.
A few days later, in a game Konyashkin-Titlyanov, Black replied
20 . . . h5!, and after 2 1 'W b 1 :rs 22 f7 + h7 23 .txe6 f6 24
.txf5 gf 25 'Wxf5 + 'itfg8 26 'Wg6 'We4 27 h6 + h8 28 f7 +,
the game had to end in perpetual check. However, the right line
was 2 1 'Wd 1 ! f6 22 'W b 1 'Wc6 23 .txe6, with numerous threats)
.

21 'irb 1 ! lt!h5 22 .l:l. xh5 .l:l. f6 23 .l:l. xh7+ g8 24 'irb3 'irc6 25 'irh3
1 -0.
19
20
lt!xe6
.tc3
w h8
But not 20 .td5? .l:l. fc8 2 1 lt!g5 h6, and Black has everything
in order.
.t6
20
After 20 . . . .l:l. f6 2 1 lt!xg7 wxg7 22 d5 lt!d8 23 d6! .l:l. c8 24 'ire3

e6 (24 . . . ed? 25 .txf6 + xf6 26 'ird4+ g5 27 .1:1. b3 and wins)
25 .txf6 + wxf6 26 'ird4+ f7 27 .l:l. d 1 , White's advantage is
indisputable.
21
'ir h6!
.l:l. g8
The decisive mistake. In Vaiser's opinion, 21 . . . .l:l. f7! 22 iDeS

'ireS 23 .txf7 be 24 .txg6 'irg8 25 .te4 leads to unclear play.
22
.l:l. el
.l:l. g7
In his analysis Vaiser has shown that on 22 . . . lt!d8, White

attains his goal with 23 lt!f8!
23
g4!
lt!a5
23 . . . g5 is no better : 24 .td3! .l:l. f7 25 .tg6 wg8 26 .txf7 +

xf7 27 d5!
24
25
.td3
.tal
'irc6
.1:1. 7
Things also go badly for Black in the endgame that results from
25 . . . 'irf3 26 g5 'irh5 27 'irxh5 gh 28 f4 .l:l. gg8 29 wf2 .tg7 30
lt!xg7 xg7 3 1 .l:l. xe7 + f8 32 l:t xh7.
.t g7
26
g5
27
28
29
d5!
.txg7 +
.txg6
1-0
'irxd5
g8
Game No. 1 2
Polovodin-Maslov
Leningrad 1990
1 d4 lt!f6 2 c4 g6 3 lt!c3 d5 4 lt!f3 .tg7 5 cd lt!xd5 6 e4 lt!xc3
7 be c5 8 l:t b1 0-0 9 .tel cd 10 cd 'ira5+ tt .td2 'ir x a 2
lt!d7
12
0-0
Over the last three years, a good many games and analyses
relevant to this move have accumulated. I shall here quote the
most important of them.

13
.tb4
The natural move, which is usually played almost automatically.
However, in Nemet-Kozul, Liechtenstein 1 989, White decided to
postpone this active bishop sortie and drive the queen away : 1 3
J:l. a 1 'ire6 ( 1 3 . . . 'ir b2 i s also playable) 1 4 'ir b 1 'irb6 1 5 'ird3 'ird8
16 l:tfcl b6 17 .tb4 ltlf6! (but not 17 . . . .tb7 18 e5!) 18 ltle5
.tb7 1 9 ltlc6 .txc6 20 .J:I. xc6 aS 21 .td2 (other bishop moves
are answered by 21 . . . b5, when White has inadequate compen
sation for the pawn) 21 . . . J:l. c8 22 .J:I. ac 1 l:l. xc6 23 l:l. xc6 'ira8 24
d5 ltld7. Black's chances are better; here is how the game ended :
25 'ifa3 .td4 26 .th6 J:l. e8 27 .tb5? ltlc5! 28 .J:I. c7 'ird8 29 .J:I. a7
'irb8 30 .txe8 'ifxa7 3 1 'irh3 'irb8 32 .tc6 ltlxe4 33 d6 ltl xd6 34
'ird7 ltlf5 35 .td2 'ire5 0-1 .
ltlb6 (40)
13
40
w
Before proceeding to the game continuation ( 1 4 ltle5), let us

look at some alternatives.
The exchange of the pawns on e4 and e7 took place in Vaiser
Kozul, Ptuj 1989: 14 .txe7 J:l. e8 1 5 J:l. a 1 (or 1 5 .tc5 l:l. xe4 1 6
.t d 3 J:l. e8 1 7 J:l. a 1 'ir b2 with a draw, but Black also has the
interesting reply 15 . . . ltld7!) 1 5 . . . 'ire6 16 .tc5 'ifxe4 1 7 ltle5
.txe5 18 .tf3 .txh2+ 19 xh2 'iff4+ 20 g1 ltld7 21 .txa7
ltlf6 22 'irc1 1Wf5 23 J:l. e 1 J:l. xe 1 + 24 'ifxe 1 'iff4. White has the
advantage of the bishop pair, which compensates for the pawn
minus.
An equal game results from 14 .tb5 .td7 1 5 .txd7 ltlxd7 1 6
.t.xe7 l:l. fe8; Riemersma - Conquest, Dordrecht 1 988. I t is worth
considering 14 d5, with the threat of 1 5 .txe7.

The game Halifman-Gavrikov, Moscow 1 988, took the follow
ing interesting course :
14 'fldJ!? r1 e8 15 liJg5!
Exploiting the unfavourable placing of the enemy queen, White
sets up the threat of 16 f1 a l . The tactical justification, given by
Halifman, is : 1 5 . . . .txd4 16 .td 1 ! f1 d8 ( 1 6 . . . .te6 1 7 'ifxd4
'ifxb 1 18 liJxe6 fe 19 .tc3, or 16 . . . 'ifc4 17 'iff3 and 1 8 .tb3) 1 7
.tb3 .txf2+ ( 1 7 . . . 'flxf2 + 1 8 w h 1 ) 1 8 h 1 f1 xd3 1 9 .txa2
and wins.
15 . . . .te6 16 d5 .i.. d 7
The attempt to take advantage of the queen on the d-file with
16 . . . f1 ad8 is no good in view of 17 de f6 1 8 'flh3 fg 19 .tb5.
17 'fl/3 f5!
The only move. 1 7 . . . .tf6 loses at once to 1 8 .txe7!, while 1 7
. . . f6 1 8 .C. a 1 'flc2 1 9 .C. fc l 'ii' b 2 20 ..tc3 'flb3 2 1 .C. cb 1 'ifc2 22
.C. b2 is also bad for Black.
18 .td3 .i../6 19 ef
Rather recklessly played. 1 9 liJe6 or 1 9 h4 was stronger, and
may have given a sharper attack.
19 . . . ..txg5 20 fg 'flxd5 21 .te4 'fl e6 22 .i.. d2!
Not 22 'flh5 h6 23 f4 'fl xe4 24 fg, which fails to the intermediate
check 24 . . . 'fle3 + .
2 2 . . . hg!
Other moves quickly lose.
23 .txg5 .tc6 24 .txc6 'flxc6 25 'ii' h3 'fl d5 26 .te3
Here White had the much stronger 26 'flh6! 'iff7 27 .te3,
winning the pawn back and retaining a positional advantage.
26 . . . liJc4 27 'flg4 g7 28 .td4+ J-J.
So 14 'fld3 gave White a dangerous attack. However, the
vulnerability of the queen on d3 was emphasised in a game
Tukmakov-Gavrikov, Moscow 1 989, which went 14 . . . .i.. e6!
(41) .
Black threatens 1 5 . . . .tc4, which Makarichev suggested
parrying with 15 liJdl .C. fd8 16 d5. But then Black would carry
out much the same decisive tactical operation as in the actual
game : 1 6 . . . .txd5! 1 7 ed liJxd5! and there is no defence against
1 8 . . . liJxb4 or 18 . . . liJf4. The correct line for White is 15 .C. at
'flc4 16 .txe7 'flxd3 1 7 .txd3 .C. fe8 1 8 .tc5 .tc4 19 .txc4
liJxc4, with a roughly equal ending; or 1 5 . . . 'flb3 1 6 .txe7 f1 fe8
17 .tc5 liJa4 18 'flxb3 .txb3 1 9 .tb5 liJxc5 20 .txe8 .C. xe8 2 1

41
w
: a3, as in S. Ivanov-Novik, USSR 1 989, and now 2 1 . . . .tc4!

22 de .txfl 23 xfl , again with an equal position.
Tukmakov played 15 d5??, not even suspecting the danger
awaiting him : 1 5 . . . lll x d5!, an attractive stroke which decides the
issue at once. After 16 ed .tf5, not only the rook on b1 but also
the bishop on e2 is left under attack, and White could very well
cease resistance. This actually happened 12 moves later.
14 lll e5 (42)
14 : a 1 "ife6 1 5 "ifc2 is interesting, but has yet to be tried out
in practice. It was Neverov who suggested jumping to e5 with the
knight, and the move was first played in a game Alterman
Rogozhenko, USSR 1 989.
42
B
14
f6
In the original game with this line, Black decided to eliminate

the knight at once, and was quickly crushed : 14
.txe5 1 5 de
"ife6 1 6 "ifd4 .td7 1 7 f4 f6 ( 1 7 . . . .tc6!? was more tenacious;
after 18 f5 "ifd7, the play isn't so clear) 18 ef "ifxf6 19 "ife3! : ac8


20 .tc5 .tc6 21 f5 'ife5 22 l:l. xb6! ab 23 .tc4+ l:l. f7 24 .txf7 + !
and White won.
The retreat of the queen to e6 in these circumstances led to
disaster in Ivanov-Maslov, Leningrad 1 990 : 14
'ife6 16 fe f5
1 6 .tc5 fe 1 7 .txb6 l S de ab 1 9 'ifd4 'ifc6 20 l:l. fc l (or 20 .tc4+
.te6 21 .txe6 + 22 l:l. xb6 'iff5 23 l:l. xb7, with the better chances)
20 . . . l:l. a4 (he should have played 20 . . . 'ifd7 2 1 'ifxb6 'ifdS, with
a complex struggle) 2 1 'ifd l ! 'ifd7 22 : xeS! and wins.
14
.te6 is also dangerous for Black: 1 5 .txe7 .: feS 1 6 .th4!
(stronger than 16 .tc5 .txe5 17 de .tc4 lS .txc4 'ifxc4, with
equal chances; Halifman-Kindermann, West Berlin l 9S9) 16 . . .
g5 ( 1 6 . . . .:t eeS may be better) 1 7 .txg5 (or 1 7 .tg3 : adS l S
.tb5 .: rs, as in Sakayev-Novik, USSR 1 990, and now 1 9 'ifh5!
f6 20 d5 fe 2 1 de 'ifxe6 22 'ifxg5 with a clear plus) 17 . . . f6 1S
.:t a l 'ifb3 ( 1 S . . . "ifb2 1 9 .!Lid3 'ifxd4 20 .te3 'ifxe4 was better) 19
'ifd2 l:l. acS 20 .td1 'ifc3 2 1 'ifxc3 l:l. xc3 22 .td2, and White wins.
. .
. . .
15
.!Lic4!
In the game Shevelev-Lagunov, USSR 1 9S9, in which 14 . . . f6

was played for the first time, White withdrew his knight to f3,
giving Black a good place for his queen : 15 .!Lif3 'iff7! 16 'if c 1
.t d 7 1 7 'if a 3 l:l. fcS! 1 S .txe7 (otherwise 1 S . . . .tfS) 1 S . . . l:l. eS
1 9 .tc5 l:l. xe4 20 l:l. xb6 (20 .txb6 l:l. xe2 is worse) 20 . . . l:l. xe2 2 1
.l:t xb7 a S 22 'ifd3 .:t eeS 2 3 d5 (43) .
43
B
Now after 23 . . . J:. abS 24 l:l. a7 a4, the game was eventually
drawn. Lagunov points out an interesting trap that Black could
have set : 23 . . . .:t eeS! 24 .ta3 (also after 24 l:l a7 l:l. xa7 25 .txa7
a4, the advantage is with Black) 24 . . . .tf5! 25 'ifb5 .td3! and
wins.

Apart from the retreat of the knight at move 1 5, practice has
seen t5 :. a t 11re6 t6 .t g4 f5 1 7 ef gf 18 .tf3 l:r. d8 19 .l: e 1 .txe5
20 .l: xe5 11rxe5 21 de .l: xd 1 + 22 .l: xd 1 f7, with a roughly equal
game; Neverov-Malishauskas, Podolsk 1 989. Or t6 d3!? 11rxe4?
( 1 6 . . . 11rf7 is safer) 1 7 .tf3 11rf5 1 8 .l: e 1 .l: e8 1 9 .tc5 .td7 20
.txb7 .l: ab8 2 1 .l: xa7 with the advantage; Agrest-Maslov, USSR
1 990.
t5
t6
:at
xc4
bl!
t6
11rb2 loses the queen to 17 .txc4+ h8 18 .l: b l !. On
the other hand, t6 . . e3 transposes into the game after 1 7 : xa2
xd l .
. . .
t7
t8
t9
.l: xal
xd t
h8
.tc4+
: xdt (44)
44
B
How should we assess this rook and bishop ending? Black is a

pawn up, and has two connected passed pawns on the queenside.
White has powerful pawns in the centre, and the more active
bishops. Overall, White's position looks the more attractive, and
indeed in the present game his conduct of the ending brings
success. (From now on, I shall make use of the winner's notes.)
t9
.td7
After t9 . . . a6 20 f3 : e8 21 .tb5 .te6 22 d5 ab 23 : xa8 : xa8

24 de .tf8 25 : d7 : a6 26 .txe7 .txe7 27 : xe7 : b6 28 : e8 +
g7 2 9 .l: b8 .l: xe6 3 0 : xb7+ (threatening .l: xb5), White would
gain a clear advantage (Almeira). But a better move, undoubtedly,
was t9 . . . : d8, clarifying the position in the centre at once : 20

J.xe7 l:l e8 21 J.c5 l:l xe4 22 l:l xa7 l:l xa7 23 J.xa7. Material
equality is restored, and White has a minimal edge.
l:l fe8
20 J.xe7
21
J.d6
The game Petursson-Almeira, Lyon 1 990, went 2 1 J.c5 b6,

and only then 22 J.d6. After 22 . . . J.e6 23 d5 J.d7 24 J.d3 f5
25 ef gf, White utilised the weakening of a6 with 26 J.a6!
21
.l:l d
J.f8 2 2 J.xf8 .a. xf8 23 .t d 5 ..i.c6 2 4 J.xc6 be 2 5
After 21
.a. a6 .a. fc8 2 6 f3, the double rook endgame is difficult for Black.
The ingenious tactical trick 21
.a. xe4 22 .td5 .te6 doesn't
tum out so well after 23 .txe4 J.xa2 24 J.xb7. A safer line for
Black is 21
.i.e6 22 d5 J.d7 23 J.d3 f5 24 ef gf.

. .
22
.i.f7
a5
22 . . . J.f8 23 J.xf8 .a. xf8 24 J.d5 J.c6 25 J.xc6 be 26 .a. a6

similarly leads to a bad ending for Black.
23
f4!
The advance of the centre pawns quickly decides the game.

23
.l:l a6
Black could resist more stubbornly with 23 . . . J.f8 (23 . . . .i.h6

24 g3) 24 .i.xf8 .l:l xf8 25 J.d5 J.c6 26 ..i.xc6 be 27 .l:l c l .l:l fd8
28 .l:l a4, although even then it isn't simple to hold the ending.
..i.f8
24
eS
25
J.d5!
A tactical resource which consolidates White's advantage. After

25 J.xf8 .l:l xf8 26 .i.d5 b5!, the black pawns would already be
much the more dangerous.
.l:l a7
25
If 25 . . . J.xd6, the intermediate move 26 J.xb7 is decisive.
Nor can Black save himself with 25 . . . .l:l xd6 26 ed b6 27 l:l b l
J.xd6 (27 . . .a. b 8 2 8 l:l xa5!) 28 .a. xb6 .i.xf4 2 9 f2 .txh2 30
.l:l xf6.
.
26
27
J.xf8
l:l xf8
e6
J.c8
Other moves likewise fail to rescue Black : 27 . . . J.c6 28 J.b3

a4 29 d5 .i.b5 30 d6!, or 27 . . . i.b5 28 l:l b2 .ta4 29 l:l a l .tc6
30 J.c4.
bS
28 l:l el
If 28 . . . b6, then 29 .tc6 l:l e7 30 d5, with a won position.
29
.tc6
l:l e7

b4
30 d5
Or 30 . . . J:[ d8 3 1 .i.xb5 : xe6 32 : xe6 .i.xe6 33 de J:[ xd 1 +
34 f2 J: d2 + 3 5 f3, and the pawn queens.
: at
31
1-0
Game No. 1 3
Gelfand-Ivanchuk
Tilburg 1990
I d4 .!Zif6 2 c4 g6 3 .!Zic3 d5 4 cd ll:lxd5 5 e4 .!Zixc3 6 be .i.g7 7
.!Zif3 c5 8 : bt 0-0 9 .i.e2 cd 10 cd 1W a5 + I I .i.d2 'lf xa2
12
0-0
The present game will feature 12

.i. g4 - we shall come to
it shortly.
In Games 1 1 and 1 2, we looked in detail at 1 2
ll:ld7 and 1 2
b6 . Since the black queen must return t o its own camp sooner
or later, the immediate 12
1We6 is also sometimes played. Let
us digress a little from our principal game and consider this
manoeuvre. White protects his pawn with 13 1W c2, giving this
position (45) :
.
. . .
45
B
The queen doesn't feel very comfortable on e6, and it is now

best to move it to d6 or c6.
Gelfand-Kindermann, Debrecen 1 989, went 13
1Wd6 14 d5
b6 ( 1 4 . . . .!Zia6 is safer) 1 5 .i.b4 1Wd8 1 6 J:[ fd 1 .!Zia6 1 7 .i.a3 ll:lc5
1 8 .!Zid4 .i.xd4 1 9 : xd4 1Wd6 20 1W c3 1Wf4 21 .i.b5 .i.d7 22 .i.xd7
ll:lxd7 23 .i.xe7 J:[ fe8. At this point, as Gelfand has shown, White
could have won beautifully with 24 d6 l:t ac8 25 J:[ c4 ! 1W xe4 (25
. . . J:[ c5 26 l:t xc5 .!Zixc5 27 1Wd4) 26 l:t xc8 1W xb 1 + 27 1W c l 1Wxcl +


28 .l h c 1 fS 29 .l:l. c7 ll:lf8 (29 . . . ll:lcS 30 .I:!. xeS!) 30 h4, etc. Instead
he played 24 g3?, and after 24 . . . 11'eS 2S -*.a3 .l:l. ac8 26 11Fd2 .l:l. c2
27 11'xc2 11'xd4 28 .l:l. e 1 ll:lcS 29 -*.xeS be 30 .l:l. d 1 11'xe4 3 1 11'xcS
11Ff3!, a level ending arose and a draw soon resulted : 32 .l:l. c l (32
11Fd4? .l:l. e4!) 32 . . . .l:l. d8 33 .l:l. e 1 .l:l. xdS (33 . . . 11FxdS? 34 .l:l. e8 + ) 34
.l:l. e8 + g7 3S 11'f8 + f6 36 11'e7 + , 1-t .
Interestingly enough, the position in diagram 4S occurred in
another game between the same players a year later (Dortmund,
1 990). This time, after 13 . . . 11Fd6 Gelfand innovated with the
immediate 14 ..i.b4!. There followed 14 . . . 11'd8 1 S dS -*.g4 ( 1 4
. . . b 6 transposes into the previous game, which should not have
turned out well for Kindermann) 1 6 .l:l. fc 1 ll:la6 17 -*.a3 -*.xf3 1 8
gf! (White has obtained a definite plus) 1 8 . . . 11'd7 1 9 -*.xa6 ba
20 11'c7 11Fh3 21 11Fg3! 11'xg3 22 hg, and Black has a miserable
position.
Gelfand-Kamsky, Til burg 1 990, went 13
11'c6 14 11Fd3 11'd6
1 S .*.b4 11'd8 16 11'a3 (of course White had no reason to exchange
queens, but then his queen didn't feel too comfortable on d3 either)
1 6 . . . ll:lc6 1 7 .i.cS (46) .
. . .
46
B
After lengthy manoeuvres ( . . . 11Fd8-aS-a2-e6-c6-d6-d8), the

black queen has returned to its 'rightful' place. Several tempi have
been lost, and in addition White is threatening d4-dS. However,
Black found an interesting tactical resource : 1 7 . . . .*.xd4 !? 1 8
ll:lxd4 lll xd4 1 9 ..i.xe7 ll:lxe2 + 20 h 1 11'e8 2 1 .*.xf8 11'xf8 22
11'e3 aS (while White surrounds the knight, the black a-pawn
speeds forward) 23 11'xe2 a4 24 .l:l. bS 11'e7 2S .l:l. a 1 .*.e6 26 eS a3
27 11Ff3 a2 28 11'xb7 11Fd8 (Black is the exchange down, but his far
advanced a-pawn enables him to save himself) 29 h3 .l:l. c8 30 .l:l. b2

'ifgS 3 1 J:t bxa2 (forcing the issue) 3 1 . . . J:t c l + 32 J:t xc l 'ifxc l +
33 h2 .txa2 34 'ifaS + g7 3S 'ifxa2 'iff4+ 36 g3 t - t .
And now, one more digression : the new move 1 2
aS!? was
played in Hebert-Kozul, Toronto 1990. No other examples are
available, but in this case the advance of the rook's pawn brought
Black success : 1 3 .tgS a4 14 .tbS (better is 14 'ird3) 14 . . . .td7
IS .txd7 l0xd7 16 J: xb7 'ife6 17 dS 'ifxe4 IS J: xd7 a3 19 : xe7
'ir b4 20 : ee l a2 21 .te7 'irbS 22 'ir d2? (the right course was 22
.txf8 a l ('if) 23 'ifxa l : xa l 24 .txg7 : xe l 2S : xe l xg7 26
h4! 'irxdS 27 J:t e3, with equality) 22 . . . : res 23 d6 a l ('if ) 24 : xa l
: xa l 2S : xa l .txa l 2 6 h3 'ird7, and Black won.
12
.t g4
Black leaves his pawn on b7 undefended, but takes aim at the
white d-pawn. 13 J: xb7 .txf3 14 .txf3 .txd4 I S eS l0a6 16
J:t xe7 : adS leads to sharp play, with material equality.
h6
13
.t gS!
In the game Sakayev-Bukhman, USSR 1 9S9, Black retreated
with 1 3 . . . 'ife6, and there followed : 14 dS (at this point, taking
the b-pawn doesn't pay at all : 1 4 : xb7 'ifxe4 I S : xe7 'irdS, with
advantage to Black) 14 . . . 'ifxe4 (or 14 . . . 'ird7 I S 'irb3 b6 1 6
'ifa3, and White has the initiative) I S 'ird2! f6 (the preliminary I S
. . . a S was better: 1 6 J:t xb7 f6 1 7 .te3, with compensation for the
pawn) 16 .te3 aS 1 7 : fc l ! l:t cS 1 S h3 .td7 19 J:t xcS+ .txcS
20 .td3 'if a4 (20 . . . 'irxdS loses to 21 J:t c l 'ird7 22 .tc4+ hS
23 .te6! 'ifxe6 24 'irdS+ 'irgS 2S : xeS) 2 1 .tcS 'ird7 22 'ife2
.tfS 23 .tc4 h8 24 : b6 l0a6 2S .td4 l0c7 26 lOeS! fe. White
could now have won immediately with 27 'ifxeS + gS 2S d6+
e6 29 de .
14
.t e3
l0c6
Ivanchuk twice reached this position in the Manila Interzonal,
1 990. Gurevich-Ivanchuk continued : 14 . . . b6 l S h3 .txf3 1 6
.txf3 e S 1 7 dS l0 d 7 I S : e 1 J:t fdS 19 'ir d 3 lOcS 20 'irbS .tfS 2 1
: ed 1 l0b7 2 2 J: a l t -t .
. . .
IS
dS
lOaS
In the game between the same opponents in Manila, Black

preferred I S . . . lOeS, and after 1 6 : xb7 e6 1 7 : e l .txf3 I S gf
J:t fdS 19 d6 'ira3 20 d7 'ife7 2 1 .tbS 'irh4 22 .tfl a draw was
agreed. White has a strong passed d-pawn, but his king's residence
is weakened, which gives Black sufficient counterchances. In the
present case Ivanchuk is attempting to improve on his play, but
Gelfand succeeds in exploiting the position of the knight on the

edge of the board.
16
17
.tc5
e5!
.tf6
The positional sacrifice of a second pawn.

17
18
19
h3
.txf3
.txe5
.t xf3
lUeS
In a detailed examination of the game, Gelfand and his trainer

Kapengut indicate the stronger 19 . . . .l:t ae8 20 d6 (20 :t e l lll c4
2 1 .l:t xb7 .*.d6 22 .l:t xa7 11Vd2 leads to equality) 20 . . . ed 2 1 .td5
lll c4 22 11Vg4 (the tempting 22 .l:t xb7 de 23 11Vf3 lll d 2 24 11Vd3 comes
up against 24 . . . .th2 + ! 25 xh2 lll xfl + 26 gl 11V a l ! lvanchuk; but 22 .tb4! is worth considering) 22 . . . de 23 11Vxg6+
(if 23 .txc4, Black can defend his g-pawn with 23 . . . 11V c2) 23 . . .
.*.g7 24 .l:t xb7 lll e 5! 25 11Vxg7 + xg7 26 J.xa2 .l:t b8 27 .l:t xa7
: a8, with equality.
20
d6!
Now it turns out that Black cannot take the d-pawn, so he has
to part with the exchange.
20
11Vc4
The endgame which now arises is difficult for Black, but the
alternatives are thoroughly bad: 20 . . . ed 21 .td5 ll\c4 22 .l:t xb7,
or 20 . . . 11Ve6 21 d7 .l:t ed8 22 11Va4 lll c4 23 .txb7 .l:t ab8 24 .txa7.
21
22
d7
de(11V)+
11Vxc5
.l:t xe8
For the moment Black is quite well off from the material point
of view, with three pawns - more than the equivalent - for the
exchange. His pieces, however, are badly co-ordinated, and this
leads to more losses.
23
11Va4
23 11Vd7 looks tempting, with the aim of exploiting the exposure

of the black king after 23 . . . 11Vc8 24 11Vb5 J.c7 25 .l:t fc l .l:t d8 26
1t'c5 lll c 6 27 .txc6 .td6 28 11Ve3!. But the raid with the queen
would be refuted by 23 . . . f8! 24 .l:t b5 11Vc7.
23
24
lll c6
.l:t xb7
After 24 :t fc l 11V d4 25 .l:t c4 11Vd7 26 .l:t xc6 be 27 .txc6 11Vf5! 28

J.e4 11V c8, it is Black who has the better game.
24
.l:t c8

25
-*.xc6
25 .l: b5 11rc3! 26 .i.xc6 11r xc6 is insufficient to win.

25
26
: xc6
.J: xe7
Already the advantage in material is with White.

26
11rxe7 27 11rxc6 -*.d4 28 11rd5 -*.b6 29 g3 b5 30 gl
g7 31 .J: bl 11re2 32 11rf3 11re6

White's advantage is indisputable, but to convert it to a win he

needs to exchange queens.
33 11rc3 + h7 34 .l: el 11rd5 + 35 11rf3 11rd7 36 .J: d l 11re7 37
.l: d3 g7 38 11rc6 11re2 39 l:t3 11ra2? 40 11rc3+ g8 41 11rb3
White's main task is accomplished; the rest is simpler.
41
11r xb3 42 : xb3 f8 43 f3 e7 44
d6 46 : dl + we6 47 4 6 48 : c2

: bl
e6 45 e4
The rook cannot be prevented from penetrating the Black

position.
48
-*.d4 49 e4 .tb6 50 .l:l. c6 + e7 51 f4 d7 52 .1:1. (6
e7 53 .l:l. c6 d7 54 d5 -*.c7 55 .l:l. c4 aS 56 .l:l. c6 -*.d8 57 .l:l. a6
e7 58 e4 .tc7 59 dS .td8 60 .l:l. a7 + f6 (47)
. . .
47
w
61
J: a8!
Now if 61 . . . .tc7, White

trapped. If 61 . . . e7, then
decides the game; once the
exhausted, Black must move
61
62
wins with 62 c6, as the bishop is

62 c6, and an original zugzwang
possible kingside pawn moves are
his a-pawn for lack of anything else.
-*.e7
.l:l. xaS
1-0
White cannot be stopped from approaching the f-pawn with his

king and then winning it with the aid of the rook, after which
further resistance is senseless.

Game No. 1 4
Karpov-Kasparov
World Championship (15th game)

Lyon 1990
Before proceeding with this game, I should explain that nearly
all the material for the present book was assembled before my
last match - the fifth - with Kasparov; and that when I began
putting the manuscript in order immediately the match was over,
I allotted two places to the four Griinfeld games played in New
York and Lyon. The point is that the ninth game of the match
was spoilt by my blunder in the ending, while in the thirteenth I
never managed to extract much from the opening; so I decided
not to include either of them among the principal (numbered)
games. The fifteenth and seventeenth match games were a different
matter - they were assigned to places 14 and 1 5. But once I set
about annotating the first of these games - with my impressions
of it still fresh - I unexpectedly realised that it contained a far
greater number of interesting subtleties and unusual situations
than might at first have been supposed. Immersing myself in the
analysis, I detected - strangely enough - more and more new
ideas and problems which had to be looked into. The result was
that whereas in other cases I had merely told the story of a game,
here I produced something more like a novel.
To be sure, by the exhaustive analysis, by the sheer abundance
of the variations given, this game stands clearly apart from all
others in the book. What is offered here for the reader's attention
is virtually the definitive theory of an important Griinfeld system
that was tested out in this match.
1
2
3
d4
c4
l0c3
l0f6
g6
d5
Ever since the World Championship Match of 1 986, the

Griinfeld Defence has been firmly incorporated into Kasparov's
repertoire, and together with the King's Indian it today constitutes
his main weapon against the closed openings. Obviously his choice
of the Griinfeld in the present match was no surprise to me.
4
cd
Essentially, this seems to me to be more logical than 4 .tf4 (to

which I turned during the 1 986 match), or 4 l0f3 .tg7 5 b3 de
6 xc4 (another popular line which was thoroughly tested both

in 1 986 and in the Seville encounter of 1 987) The merit of the
plan initiated by 4 cd is that White creates a strong pawn centre
within the shortest number of moves!
.
4
5
6
e4
be
lDxdS
lDxc3
j_g7
There was a time when it was thought essential to start

undermining the pawn centre at once with 6 . . c5. The argument
was that otherwise (after 6 . . j_g7), White could play 7 j_a3, as
recommended by Simagin, and hold up Black's flank counter
thrust for a long time. But now the bishop move to a3 is considered
fairly innocuous on account of 7 . . . lDd7 (or 7
0-0 8 lDf3 lDd7)
8 lDf3 c5.
.
. . .
j_e3
The development of forces which commences with this move is

less thoroughly explored than 7 j_c4 c5 8 lDe2, or 7 lDf3 c5 8
: b1 (7 'ili'a4+ and 7 i.b5 +, which are coming into fashion, do
not appeal to me). Together with my team of seconds I had
prepared this variation specially for the match. It must be said
that, in the main, my hopes of gaining an opening advantage with
it were not unrealised.
7
8
c5
'ii' d 2 (48)
White still has quite a wide choice of continuations here. Thus,

8 j_b5+ is playable, as are 8 lDf3 and 8 :Z. c 1 , two lines which
tend to transpose.
48
B
This is the position that formed the starting point for the
theoretical contest in the games (and analysis!) of the 1 990 match.

Kasparov endeavoured four times to equalise for the Black side,
in opposition to my efforts to develop an opening initiative. The
overall score ( + 1 = 3 to White) may be considered acceptable to
both sides, and testifies rather to the problematic character of the
variation than to the attainment of a definite verdict.
8
0-0
In the ninth game, where I tried this system for the first time,
Kasparov chose a more straightforward plan, exchanging pawns
in the centre. That game is also a very important one, so we will
now embark on a long digression to look at it in detail :
8 . . . cd 9 cd lll c6 10 r:J. dl
ECO only considers 1 0 lll f 3 .t.g4 1 1 r:J. d 1 (nor can White count
on anything after 1 1 .i.b5 0-0 12 .t.xc6 be 1 3 : c 1 .t.xf3 14 gf
1r d7) 1 1 . 0-0 1 2 .t.e2 : cs 1 3 0-0 b6 ( 1 3 . . . .t.xf3 14 .t.xf3
e5 1 5 d5 lll d4 doesn't look bad either) 14 d5 .t.xf3 15 gf lll e 5 1 6
r:J. c l 1Wd7 1 7 f4 lll c4!. After 1 S .t.xc4 1rg4+, Black is guaranteed
a draw; Haik-Granda, Dubai 01 1 9S6.
10 . 1Va5 (49)
The game Kozul-Dorfman, Marseille 1 9S9, in which the inno
vation 10 r:J. d1 was tried out, went 10
e6 1 1 lll f3 0-0 1 2 .i.b5!?
(the tempting 13 .i.h6 is parried by 1 3 . . . 1Wa5 1 4 1rxa5 lll x a5 1 5
.t.xg7 xg7 1 6 .t.d3 .t.d7 1 7 d2 : res, o r by the quite good
alternative 1 3 . . . f5!? 14 .i.xg7 xg7 1 5 e5 1rd5 16 .t.e2 b6
1 7 0-0 ib7, retaining counter-attacking possibilities in the
middlegame) 12 . . . id7 1 3 0-0 (after 13 ih6 ixh6 14 1V xh6,
the reply 14 . . . 1Wa5 + is rather risky on account of 1 5 : d2.
Similarly, 1 3 . . . lll e 5 14 lll x e5 ixb5 1 5 ixg7 xg7 is inadequate
in view of 1 6 h4. But 1 3 . . . f5 14 .t.xg7 xg7, as in the previous
note, is fully viable, since 15 e5 is met by 15 . . . lll x e5 16 de ixb5
17 1Wb4 1V b6 - the simple 17 . . . ireS is also possible - 1S lll g 5
: res 1 9 l:l. d6 ic6 20 lll x e6 + gS, with complications; or 1 S
lll d 4 r:J. ad8, with the idea of answering 1 9 1Vxb5 with 1 9 . . . : xd4!,
or 1 9 1re7+ with 19 . . . gS! 19 a4 id7, or 19 lll x b5 with 1 9
. . . r:J. xd 1 + 20 xdl a 6 2 1 1We7 + r:J. f7) 1 3 . . . a 6 ( 1 3 . . . lll e 5 is
unconvincing; although 14 de ixb5 1 5 1V b4 ireS 16 : fe l b6 is
satisfactory for Black, the unassuming 14 ie2 gives White the
better prospects) 14 ie2 1ra5 1 5 d5 1rxd2 (of course not 1 5 . . .
ic3? 1 6 de! ixd2 1 7 ixd2, with a won position) 1 6 : xd2 ed
1 7 ed lll e 7, and the ending clearly favours White. For example,
he can play to restrict the knight's mobility with 1S d6 lll f5 ( 1 S
. .
. .
. .

. . . c6 1 9 ..tb6) 1 9 ..tb6 and then g2-g4; or else 1 8 ..tc5 l::tfe 8
1 9 d6 f5 20 g4 h6 2 1 h3, with the advantage.
In Seirawan-Sokolov, Novi Sad 01 1 990, Black played the
modest 10 . . . 0-0, but after 1 1 f3 ..i.g4 1 2 ..te2 l:t cS 1 3 0-0
1rd7 14 h3, he sacrificed his bishop : 14 . . . ..th5 1 5 g4 ..txg4 1 6
h g 1rxg4+ 1 7 h2 1rxe4 1 8 g5 1rfS 1 9 f4 e 6 20 ..td3 1rf6 2 1
..te4. Black's compensation for the piece i s insufficient, and White
eventually prevailed.
Whatever the reason, Kasparov preferred to exchange queens . . .
49
w
1 1 1i'xa5 xa5 12 !3
This endgame had arisen once before, in Yusupov-Gulko,
Linares 1 989. On that occasion White preferred 12 ..td3 0-0 1 3
e2 ..td7 1 4 l:t c 1 l:t fc8 1 S d2 e6 1 6 : xeS + : xeS 1 7 l:t c 1
l:txc l 1 8 xcl fS, and the minor piece ending was quickly
drawn after 19 e2 a6 20 ..tf4 c6 2 1 ..td6+ e8 22 c3 ..tf6
23 ..tc7 ..te7 24 f3.
12 . . . 0-0
Here 12 . . . ..tg4 is premature on account of 1 3 ..tb5 + .
1 3 ..te2 ..td7 (50)
14 ..td2!
Typical of Grtinfeld endings. If the knight retreats, then after
1 4 . . . c6 1 S dS eS (in the event of 1 S . . . d4 1 6 xd4 ..txd4
1 7 ..tb4 ..tf6 1 8 f4 aS 19 ..ta3 ..tc3 + 20 f2 l:t fe8 2 1 lt b 1 , or
20 . . . ..tb4 2 1 ..txb4 ab 22 l% d2, Black has many problems) 1 6
xes ..txe5 1 7 f4 ..tc7 1 8 : c t : res ( 1 8 . . . ..tb6 1 9 ..tb4 : res
20 d2) 19 ..te3 e6 20 ..tc4, White goes ahead with his policy
of constriction. 19 f2 is also playable.

50
w
14 . . . b6 15 0-0
After pondering it for a long time, I refrained from doubling
Black's pawns with 1 5 .t.xa5 ba, for he would then acquire an
adequate initiative. However, as Zaitsev has shown, 1 5 .t.a6!? was
more promising. The analysis of this endgame points to the
conclusion that the entire variation, beginning with 8 . . . cd, fails
to secure equality for Black.
15 . 'D.fd8?!
This appears to be the root cause of his subsequent troubles. A
more natural choice seems to be 1 5 . . . .l:fc8 1 6 : c 1 (if 1 6 .t.a6,
then 1 6 . . . 'D. c2 1 7 .t.xa5 : xa2 1 8 .t.b7 'D. b8. Therefore White
should first play 1 6 .t.xa5 ba, and then 1 7 .t.a6 'D. c2 18 'D. c 1
: xa2 1 9 : c7 'D. d8, but even so, Black i s risking nothing after 20
h3 'D. a3! 2 1 : xa7 'D. xf3 22 gf .t.xd4 and 23 . . . .t.xh3) 1 6 . . . .t.g4
( 1 6 . . . lDc6 is problematic; Black aims to answer 1 7 .t.a6 with 1 7
. . . lDxd4 1 8 .t.xc8 lDxf3 + 1 9 gf .t.xc8 20 g2 .t.e6, o r if 20 : c7,
then 20 . . . .t.h3. In either case Black has definite compensation for
the exchange, but is it sufficient for equality? Thus, in the last
variation, after 21 'D. fc l e5 22 'D. b7!, the rook cannot be dislodged
from the seventh rank : 22 . . . .t.c8 23 : xa 7!) 1 7 .t.a6 (on 1 7 d5,
Black has 1 7 . . . .t.b2, when neither 1 8 'D. b1 .t.xf3 19 gf 'D. c2 20
'D. fd 1 .t.c3 nor 1 8 : xc8 'D. xc8 1 9 'D. b 1 .t.xf3 20 gf 'D. c2 21 'D. d l
.t. c 1 presents him with any danger) 1 7 . . . : xc 1 1 8 'A xe l .t.xf3
19 gf .t.xd4 20 'D. c7. White regains the sacrificed pawn, but cannot
achieve more than that.
16 : ci .t.g4 (51)
Again a straightforward decision. Perhaps it was worth prepar
ing an exchanging operation with 1 6 . . . a6!? 1 7 'D. b l ( 1 7 : c7
.t.b5 1 8 .t.xb5 ab 1 9 d5 lDc4 20 .t.b4 .t.f6 2 1 .t.xe7 .t.xe7 22
.

: xe7 : xa2 leads to complications, since 23 10d4 b4 24 10c6? b3
2S 10xd8 b2 looks too hazardous for White, while 23 : b 1 fS!? 24
lOgS fe 2S 10xe4 10d2 should lead to a draw) 1 7 . . . .tbS 1 8 .txbS
ab 19 : xbS 10c4 20 .tb4 : xa2 21 : c 1?! .txd4! 22 : xc4 .txf2+
23 fl : d 1 + 24 10e 1 .txe 1 2S .txe1 : aa l . Nor does 1 7 .tf4 pro
duce the expected effect: 17 . . . .tbS 1 8 .txbS ab 19 .tc7 :t d7 20
.txb6 10c4. Or if 1 7 .tgS, then 17 . . . h6! 1 8 ..txe7 : es, and
Black restores the balance. However, it is worth considering 1 7
dS!? .tbS 1 8 : fe 1 : ac8, with possibilities for both sides.
51
w
1 7 d5 10b7
Devices such as 1 7 . . . e6 1 8 h3 .txf3 ( 1 8 . . . ed 1 9 hg de 20
.txaS ef 2 1 .txf3 ba 22 .txa8, and Black has a lost position) 1 9
.txf3 e d 2 0 e d fail t o help Black - h e i s still faced with the same
inferior ending.
18 h3
White would be wasting time with 1 8 .tb4 e6, while 1 8 : c7?!
is a mistake owing to 1 8 . . . tOeS! ( 1 8 . . . .l:t d7 19 : fc 1 , with the
initiative) 19 : xe7?! .tf6 20 : c7 10xe4.
18 .
.txf3
Nor is 1 8 . . . .td7 1 9 .tb4 e6 20 .ta6 any better for Black.
19 .tx/3 10c5 20 .te3 : ac8 21 .tg4! :t b8
Or 2 1 . . . : c7 22 .tf4 :t b7 23 :t fd 1 , and White has a large
plus, since 23 . . . 10xe4? loses at once to 24 : cS : xeS 2S .txc8.
White also retains a substantial advantage with 22 f4 (with a view
to 22 . . . hS 23 .tf3, and the pawns creep forward) 22 . . . fS!? 23
ef gf (23 . . . .l:t xdS 24 fg hg 2S .l:t fd 1 .l:t xd 1 + is also a grim outlook
for Black) 24 .txfS : xdS 25 : xeS! : dxcS (if 2S . . . : exeS, then
26 ..te6+ f8 27 .txcS : xeS 28 :t d 1 !, and Black is in a difficult
. .

position after either 28 . . . .tb2 29 .l: d8 + g7 30 g4 f6 3 1
.tb3, o r 2 8 . . . .ih6 2 9 .l: d8 + g7 3 0 g3! f6 3 1 .tb3) 26
.te6 + f8 27 .l: d 1 e8 (27 . . . .l: c8?? fails to 28 .txcS, and 27
. . . .tb2 28 .txcS be 29 .l: d8 + g7 30 g4 f6 3 1 .tc4 brings
Black no relief either) 28 .txcS be 29 .1: b 1 ! (astonishing domination
with full material equality) 29 . . . .td4+ 30 h2 f8 3 1 .l: b8 +
<t;g7 32 g4 <t;f6 (he can't save himself with 32 . . . .l: c6 33 .l: g8 +
f6 3 4 .tdS .l: d6 3S gS + fs 36 : f8 + g6 3 7 .te4+ g7
38 : e8 .l: e6 39 .tdS .l: e2 + 40 <t;g3 hS 41 h4, or 40 . . . h6?
41 .l: g8 + h7 42 g6 mate) 33 .ic4 e6 34 .l: e8 .l: e7 3S gS + f7
36 .l: c8 g6 (36 . . . g7? 37 fS!, or 36 . . . .te3 37 g3) 37 g3
h6 38 h4 hg 39 hg eS 40 : c6+ cJi;g7 41 fS, with a simple win.
The reader will agree that these variations are pretty, although
rather long. But now to return to the game, i.e. the 9th match
game :
22 .l: c4
A mistake would be 22 .txcS be 23 .1: xeS .1: b2 24 .1: aS .id4,
when White's material advantage is neutralised by Black's pressure
against f2.
22 . . . h5
A difficult ending results from 22 . . . lDxe4 23 .l: xe4 fS 24 .txfS!
gf 2S .l: xe7, or 22 . . . fS 23 .txcS be 24 ef gf (not 24 . . . .l: xdS? 2S
f6!) 2S .txfS .id4 26 .te6+ g7 27 h2 (27 .l: a4 .l: b7 28
.l: d 1?! is less convincing in view of 28 . . . .l: b2 29 .l: xa7 .txf2+
30 h2 .th4, when neither of the alternatives bring a clear result :
3 1 d6 .tf6! 32 d7 - not 32 de?? .teS + 33 g1 .l: xd 1 mate - 32
. . . .teS + 33 cJ;g1 f6 34 ..tg4 e6 3S : a6 .td4+ , or 31 .l: fl .tf6
32 .l: xe7+ .txe7 33 .l: f7 + g6 34 .l: xe7 .l: xa2 3S d6 .l: a6! 36
d7 .l: xe6 37 .l: xe6+ f7 38 .l: h6 g7 39 .l: d6 c4) 27 . . . .l: b2 28
.l: a4 .l: b7 29 f4, with a technically won position for White.
23 .t/3 e6 24 'IJ. dl ed 25 ed
Stronger than 2S .l: xdS .l: e8!, when Black is defending actively.
25 . . . .ie5 (52)
At first sight it seems that on blocking the d6-square, Black will
have constructed an almost impregnable fortress.
26 g4!
This breaking-up of the pawn chain dispels any illusions of
equality. The endgame was and still is unattractive for Black.
26 . . . hg 27 hg lDb7?!
Perhaps 27 . . . .1: bc8 or 27 . . . .1: b7 was more logical, but even

52
w
so, by deploying his forces in the same way as in the game g4-g5, ..;. g l -g2, and the transfer of the rooks to the h-file - White
would clearly have the better chances.
28 TJ. a4
For good measure, Black's knight is now driven onto a bad
square, since 28 . . . TJ. a8 fails to 29 d6.
28 . . li:Ja5 29 g5!?
It is also worth considering 29 TJ. e4!?
29 . . . TJ. bc8 30 J:.e2 J:.d6
If 30 . . . TJ. c2, a good reply is 3 1 J:.d3 TJ. c3 (3 1 . . . TJ. b2? 32 J:.c1,
and the rook is trapped) 32 J:.a6, after which White again carries
out the plan beginning with g 1 g2
31 g2!?
More accurate than 3 1 J:.a6 TJ. c2 32 ..;.g2, which could be met
by 32 . . J:.c5 33 J:.d3 TJ. c3; whereas now, 3 1
TJ. c2 32 J:.d3
would immediately put Black in a critical position.
3 1 . J:.c5 (53)
.
. . .
. .
53
w

32 J.d2??
An absurdity! I make the second move of the variation 32 J.f4
J.d6 33 J.d2, which would give White somewhat the better
chances. If instead 33 J.c1 J.fS 34 J.b2 : cS ! (bl,lt not 34 . . .
J.g7 35 J.f6!, when 35 . . . J.xf6 36 gf : d6 37 : h4 : xf6 3S
.J:I. dh 1 , or 37 . . . g5 3S .J:I. h6, leads to catastrophe) 35 .J:I. h4 J.g7 36
J.f6 .J:I. dxd5! 37 .l:l. dh 1 .J:I. xg5 + ! 3S J.xg5 : xg5 + , the play is quite
sharp. White's best line is 32 J.f4 J.d6 33 J.xd6 (33 J.a6 : c7
brings White to a dead end) 33 . . . : xd6 34 .J:I. h4 : cS (34 . . . : c2
loses to 35 : dh 1 fS 36 : hS + e7 37 : e 1 d7 3S J.g4+
c7 39 : cS + ) 35 .l:t dh 1 fS 36 : hS + e7 37 J.f3, with a
considerable endgame advantage.
32 . . . : xd5 33 J.f3 : dd8 34 J.xa5 i-J.
It is not without good reason that we have so thoroughly
examined the first duel with this variation. It is useful for the
reader to understand that the assessment of the endgame plays a
key role in this particular Griinfeld line.
Instead of S . . . 0-0 or S . . . cd, Black frequently chooses 8
"if aS
9 .J:I. bl cd (the prophylactic 9 . . . a6, defending against .J:I. b5, can
be met by 1 0 : c 1 !?. White sets up the positional threat of
d4-d5 - since the c3 point is defended! - and compels his opponent
to go into an ending with 10 . . . cd 1 1 cd "ifxd2+ 1 2 xd2. This
ending is even more favourable to White, since the point b6 in
Black's camp has been weakened. Epishin-Gutman, Frankfurt
1 990, saw instead 1 0 . . . li'ld7 1 1 li'lf3 0-0 1 2 J.d3 li'lb6 1 3 J.h6!
cd 14 J.xg7 xg7 1 5 cd "ifxd2+ 16 xd2, and White had the
advantage). The game Kozul-Polajzer, Ptuj 1 9S9, continued : 10
cd "if xd2 + 11 xd2 0-0 12 li'lf3 e6 ( more precise than the
immediate 1 2 . . . : dS 1 3 J.d3 li'lc6 14 d5 li'la5 1 5 e2 b6 1 6
.J:I. hc 1 e6 1 7 J.g5 f6 1 S J.d2, when White's characteristic bishop
manoeuvre to shake up the black position brings tangible results)
13 J.d3 lLic6 14 : be t .l:t d8 15 eS f6!? ( 1 5 . . . f5 1 6 h4 h6 1 7 e2
J.fS 1 S g3 lLib4 19 J.c4 b6 20 li'le 1 ! is less effective for Black;
Damjanovic-Kozul, Belgrade 1 9S9. Also 1 S .J:I. c4!?, recommended
by Zaitsev, deserves serious attention. Black cannot develop his
queenside freely : 1S . . . li'le7 19 .J:I. g 1 ! li'ld5 20 g4 b5 21 : xeS : axeS
22 gf ef 23 J.xf5, and White has more than enough for the
exchange) 16 e2 fe 1 7 de li'lxe5 1 S li'lxe5 J.xe5 19 J.e4 : bS
20 : cs J.d6?! (according to Informator 4 7, 20 . . . J.d4 2 1 : as
e5 22 J.xd4 .J:I. xd4 23 f3 gives White a big advantage, but this
. .

verdict is obviously exaggerated. After the simple 23 . . . ..te6 24
.l:t xa7 ..tc4+ 25 e3 b5 26 f4 .l:t bd8, threatening 27 . . . .l:t xe4+
27 xe4 ..td3 +, Black is out of danger. If instead 24 . . .1:1. bd8
25 .1:1. bxb7! .l:t d2 26 e3, the position remains unclear; neither 26
. . . ..tc4 27 f4!, nor 26 . . . .l:t xa2 27 .l:t g7 + f8 28 .l:t xh7, nor 26
. . . .l:t xg2 27 .l:t g7 + f8 28 .l:t xh7 gives Black a cast iron draw)
21 l:l. a5 a6 22 ..tb6 .l:t d7 23 h4! ? ..tc7 24 e3 ..txb6 + .l:t xb6,
and White succeeded in realising his positional advantage.
We can see that this result was due to the somewhat unfortunate
20 . . . ..td6?!. However, at move 1 6, White had an opportunity
to strengthen his play : 16 ef (the difference between 1 5 . . . f6 and
1 5 . . . f5 now disappears) 16 . . . ..txf6 1 7 e2 il:lxd4 + (the attempt
to develop with 1 7 . . . e5 leads to a worse position : 1 8 de il:lxe5 not 1 8 . . . .txe5? 1 9 .l:t xc6 and wins - 1 9 il:lxe5 .txe5 20 .te4
.l:t b8 21 ..txa7! .l:t a8 22 .tb6 .l:t xa2+ 23 el .l:t e8 24 .td5 +
.te6 25 .l:t c8!) 1 8 il:lxd4 ..txd4 1 9 .txd4 .l:t xd4, and now White
has the better chances after either 20 e3 e5 21 .l:t b5, or 20 .l:t c7
b6 (20 . . . e5 2 1 .1:1. be l .l:t d8 22 .l:t e7) 2 1 .1:1. be l .td7 22 e3 e5 23
f4 ! .l:t d8 24 .tc4+ h8 25 fe.
Evidently it is due to considerations of this kind that Black has
recently been giving preference to the move-order starting with 8
. 0-0.
9 il:lf3
Earlier, one of the main lines here was 9 d5, but then it was
discovered that Black obtains a fine game with 9 . . . e6!, for
example: 1 0 .txc5?! 'iWc7! 1 1 .td4 .txd4 1 2 cd ed 1 3 e5 il:lc6 1 4
.te2 f6 1 5 ef .1:1. xf6 1 6 il:lf3 .tg4 1 7 0-0 .1:1. e8, and already Black
has the initiative. Perhaps an even more accurate method is 1 1 . . .
ed!? 1 2 .txg7 xg7 1 3 ed .l:t e8 + 1 4 .te2 .tg4 1 5 'iWd4+ g8
16 f3 ( 1 6 d6 'iWc6) 16 . . . .tf5, and it is obvious that Black's
initiative compensates for the sacrificed pawn.
9 .l:tcl leads to the same kind of position as in the game, if
Black replies 9
'iWa5. It acquires original overtones only in the
case of 9
lll d7 (54) .
There can follow :
10 ..td3!?
Better than 1 0 d5 il:lf6 1 1 f3 ( 1 1 .td3 b6 1 2 c4 ll:lg4 1 3 .tf4
e5!) 1 1 . . . e6 (after 1 1 . . . b6 1 2 .tb5 il:le8 1 3 .tc6 .l:t b8 1 4 .tf4?!
ll:ld6 1 5 e5 il:lc4 1 6 'iWe2 il:la5 1 7 e6 il:lxc6 1 8 de .1:t a8 1 9 ef+ .l:t xf7
Black obtains a good position, but the correct line for White is
.
. . .
. . .

54
w
1 3 .i.xe8!? .l:l. xe8 1 4 ll'le2 e6 1 5 c4 ed 1 6 .i.a6 1 7 f2 1W h4 + 1 8

g 3 1W h 3 1 9 ll'lf4 1Wd7 2 0 .l:l. he l . Now, a s Vaiser demonstrates, Black
should play 20 . . . .l:l. xe3! 21 xe3! id4 + 22 e2 g5 23 ll'ld3,
with complications. In Vaiser-Romanishin, Sochi 1983, Black
neglected this opportunity and ended up in a difficult position
after 20 . . . .l:l. e7 2 1 g2 .l:l. ae8 22 d6! .l:l. e5 23 ll'ld5) 1 2 c4 .l:l. e8 1 3
.!Oe2 b6 1 4 ll'lc3 ( 1 4 e 5 .!Oxd5!) 1 4 . . . .i.a6 1 5 ie2 (or 1 5 e 5 ll'ld7
16 f4 f6, and White can't maintain his centre) 15 . . . .!Od7 16 0-0
.!Oe5 1 7 lll d 1 ed 1 8 cd ixe2 1 9 'ifxe2 f5! and Black firmly seizes
the initiative; Spassov-Tseshkovsky, Moscow 1 985.
10 . . . e5!
This pawn-break is the point of Black's whole set-up. If now 1 1
d5, Black plays 1 1 . . . f5! and obtains strong counterplay.
1 1 ll'lf3 ed 12 cd cd 13 ll'lxd4
An interesting alternative is 1 3 ixd4!? ixd4 ( 1 3 . . . ll'lf6 1 4
h 3 .l:l. e8 1 5 0-0 ll'lxe4 1 6 ixe4 ixd4 1 7 .i.xg6 ixf2+ 1 8 'ifxf2
hg 1 9 ll'le5 leads to a double-edged position) 14 ll'lxd4 lll f6 1 5
i b 1 'it"b6 1 6 0-0 .l:l. d8 1 7 .l:l. fd 1 ig4 f3 .l:l. d7 1 9 fg .l:l. ad8 20 .l:l. c4
ll'lxg4, with unclear play (Yusupov).
13 . . . ll'le5 14 ie2 'it"h4!?
Or 14 . . . ll'lg4 1 5 ixg4 .i.xg4 1 6 0-0, which is also good for
Black.
15 0-0 1W xe4 16 lbb5 ie6 1 7 .1:1./dl! .i./6 18 lll d6 'it"h4 19
i/4 .1:1./dB 20 'it"e3 ll'lg4 21 'tW/3 ie5!? 22 ixe5 ll'lxe5 23 'it" xb 7
ll'lg4 2 4 .i.xg4 'if xg4
The chances are equal; Yusupov-Timman, Belgrade 1 989.
These fairly authoritative examples show that in the current
state of theory, 8 1Wd2 0-0 9 .l:l. c 1 is inadequate to give White an
advantage.

9
'iF aS
Iri the seventeenth match game, Kasparov refrained from the
queen excursion to aS, preferring 9
.tg4 (see Game No. 1 5).
I should also mention the variation 9
cd 10 cd .tg4 l t : c1
.txf3 1 2 gf e6 1 3 .tb5 ll:lc6 (the pawn sacrifice is justified, since
after 1 3 . . . a6 14 .ta4 it is not simple for Black to complete his
development) 14 .txc6 be 1 5 : xc6 11Vd7! 1 6 : c4 (or 1 6 : a6 11Vb5
17 'iFe2 11Vb1 +, and White achieves nothing) 16 . . . f5! (precisely
calculated - 17 0-0 fe 1 8 fe e5!, and to avoid coming under attack
White has to permit a draw with 19 d5! 11Vg4+ 20 h 1 11Vf3 +
etc.) 1 7 d5 fe 1 8 fe ed 1 9 11Vxd 5 + 11Vxd5 + 20 ed : fd8, and Black
has a satisfactory endgame, as 2 1 : cs can be answered by 2 1 . . .
.tf8 ! 22 : bS a6; Vilela-Armas, Bayamo 1 988. White might
consider the prospects offered by II : b l !? or II ll:lg5!, on the
lines of Game No. 1 5.
10
:ci
e6 (55)
Another popular line here is the transition to an ending with
10 . . . cd 1 1 cd 11Vxd2 + . Compared with the variation 7 ll:lf3 c5 8
: b l , the rook on c 1 is more mobile, which increases White's
possibilities. Play might continue :
12 xd2
More precise than capturing with the knight. Let us take a close
look at the latter: 1 2 ll:lxd2 e6 1 3 ll:lb3 (after 1 3 .tbS .td7! 14
.txd7 ll:lxd7 1 5 e2 : fc8 1 6 ll:lc4 .tf8 17 .tf4 : c6 1 8 ll:la5?!
: a6 19 ll:lxb7 : xa2 + 20 f3 e5! 21 de : e8 22 : a t ! : b2! White
had some difficulties to overcome in Hiibner-Adorjan, Bad
Lauterberg 1 989) 1 3 . . . : d8 14 .tgS f6 1 5 .te3 .tf8 (or 1 5 . . .
f5 1 6 ef gf 1 7 .t b5 lhd7 1 8 0-0 ll:lf6 1 9 .tg5 .td7 20 .tc4 b6
21 lhd2 : e8 22 ll:lf3 .tc6 23 llJeS .tdS 24 .tbS : ec8 25 a4 ll:le4
26 .tf4 .tf8 27 f3, with advantage to White; Karpov-Hiibner,
Tilburg 1 980) 16 .tc4 llJc6 1 7 e2 .td7 1 8 : hd l g7 1 9 f3
: ac8 20 d5 ed 2 1 .txd5 .te8 22 ll:lc5 .txc5 23 .txc5 .tf7, and
the game heads towards a draw; Johansen-Hart, Martigny 1 986.
12 . : dB
Another well-known continuation is 1 2 . . . lhc6.
13 .tb5 .tg4 14 : c7 lhc6 15 d5 : ab8!
Black does badly with 1 5 . . . e6? 1 6 tLl gS! ed 1 7 llJxf7 : dc8 1 8
: xc8 + : xc8 1 9 ll:ld6 : c7 20 : c l ! d e 2 1 ll:lxb7 : d7 + 2 2 e 1
: xb7 23 .txc6 : e7 24 .td5 + .te6 25 .tgS! .txd5 (25 . . . : e8
26 : c8 !) 26 .txe7 .txa2 27 : c8 + , and White's win is not in
doubt; Ftacnik-Smejkal, TrenCianske Teplice 1 979.
.
. .
. .

16 f4!
In Mikh. Tseitlin-- Schmidt, Lodz 1 980, White played the
inadequate 1 6 ..t.xc6 be 1 7 : xc6 e6 18 d6 .l: b2 + 19 d3 : xa2
20 .l: hc l (or 20 h3 xf3 2 1 gf e5, and Black has no problems)
20 . . . e5! 2 1 l:tlc2 ..t.d7! 22 J:l. 6c5 J:l. xc2 23 xc2 f6, with the
more pleasant position for Black.
16 . . . e6 1 7 ltlg5 ed 18 ltlxf7 .I: deB!? 19 ed :Z. xc7 20 ..i.xc7 .l: c8
21 de .l: xc 7 22 ..i.c4!? f8 23 ltld8 be 24 ltle6+ xe6
Again, the draw cannot be avoided.
In practice, as a rule, Black nonetheless steers clear of this
ending, since White does after all dictate the play.
55
w
11
..i.h6!
A new, and quite effective, idea of Zaitsev's. By exchanging
bishops, White attempts to exploit the weakening of the dark
squares in his opponent's camp. I proceeded less convincingly in
the 1 3th game of the match, in which the diagrammed position
arose by a different move-order (9 .l: c l 1Wa5 10 ltlf3 e6). Now is
the best time to pause and examine that game.
l l d5
Practice has also seen 1 1 ..i.e2 cd 1 2 cd 1Wxd2+ 1 3 ltl xd2 ltlc6
1 4 d5?! ed 1 5 ed ltld4, with equality; or 1 1 d3!? .l:t d8 1 2 ..t.g5
f6 13 ..t.e3 cd 14 cd 1W xd2 + 15 xd2 ltlc6 16 .l:t c4 ..t.d7 17 : b l
..i.f8 1 8 e2 b6 1 9 ..i.d2 : ac8, with unclear play; Blees
Mikhalchishin, Budapest 1 990.
11 . . . ed 12 ed .l: e8!?
A game Lputian-Tukmakov, USSR 1 989, went 1 2 . . . ..t.f5 ( 1 2
. . . ltld7 1 3 ..i.d3 b5! 1 4 c4 1Wxd2+ 1 5 ltlxd2 be 1 6 ..t.xc4 should
have led to equality in Ftacnik-Hartston, Skara 1980 : 16 . . . ..t.b7
___

1 7 0-0 lll e 5. But after 1 3 c4!? 1fxd2+ 1 4 lll x d2 b6 1 5 .i.e2, White
has a minimal edge) 1 3 .i.e2 lll a6!? 14 0-0 ll. ad8 1 5 .i.xa6 1f xa6
16 .*.xeS ll. fe8 17 c4 ll. c8 18 ..id4 ll. xc4, with equality.
13 .i.e2 .i./5 14 0-0 lll d7 15 h3
A useful prophylactic move. Instead, Piket-Korchnoi, Wijk aan
Zee 1 990, went 1 5 1fb2 lll f6 ( 1 5 . . . lll b 6 is also sufficient for
equality; Seirawan-Olafsson, Novi Sad 01 1 990, continued 1 6
1f b 3 c4 1 7 ..ixc4 ll. xe3 1 8 fe 1f c 5 1 9 .i. b 5 a 6 2 0 ..ie2 lll x d5 2 1
1fxb7 ll. e8 2 2 ..ixa6 .i.xc3 !- !) 1 6 .i.c4?! ( 1 6 1fxb7 i s more
logical : 1 6 . . . 1fxa2 1 7 .i.b5 ll. ab8 1 8 1f a6 1fxa6 1 9 .i.xa6 lll x d5
20 ..ixc5 lll x c3 2 1 lll d4, maintaining the balance) 16 . . . 1fb6! 1 7
1fa3 .i.f8 1 8 ll. cd 1 .i.d6 1 9 h 3 a 6 2 0 .i.h6 1fc7 2 1 1f c 1 1fd7 22
..if4 ..ie4 23 lll g 5 b5 24 ..ixd6 1f xd6 25 lll xe4 lll xe4 26 ..id3 c4
27 ..txe4 II. xe4, with advantage to Black in the queen and rook
ending.
15 . . . lll b6
The prophylactic reply 15
h5!? is also worth considering. In
reply, 16 c4?! is dubious : 1 6 . . . 1fxd2 17 lll x d2 ..id4! 1 8 .i.xd4
ll. xe2 1 9 ..ie3 ll. e8, with the highly unpleasant threat of 20 . . .
.: 8xe3. Therefore White would have to content himself with
exchanging bishops : 1 6 ..id3 ..ixd3 17 1fxd3, which allows 1 7
. . . 1f xa2 1 8 1fb5 lll f6 1 9 c4 b6 20 rHd 1 with complex play.
After 15
lll f6 16 c4 1f xd2 1 7 lll x d2 lll e4 1 8 lll xe4 ll. xe4,
White would still retain a microscopic plus.
16 g4 ..id7 17 c4 1fxd2 18 lll xd2 lll a4
If 1 8 . . . ..id4, then 19 .i.xd4 II. xe2 20 .i.e3 f5 2 1 II. fe 1 II. xe 1 +
22 .: xe 1 fg 23 hg .i.xg4 24 ..ixc5, and White's game is somewhat
preferable.
19 ..i/3 lll c3 (56)
The more restrained 1 9 . . . b6 is also playable.
. .

20 .J:l xc3!? .i.xc3 21 li:::J e4 xe4!
An opportune counter-stroke. After 2 1 . . . .i. g7 (2 1 . . . ..td4?
22 ..txd4 cd 23 li:::J f6+ and wins) 22 li:::J x c5, with a pawn for the
exchange, White would retain a powerful initiative.
22 ..txe4 .J:l e8 23 J.. d3 b6 24 g2 /5!? 25 gf ..tx/5 26 ..txf5
gf 27 : dl
Holmov recommends 27 .J:l c l ..tf6 (27 . . . ..td4? 28 ..txd4 cd
29 .J:l d l .J:l e4 30 f3, and endgame difficulties arise for Black) 28
f3 .J:l e4 29 .i.f4, and the bishop threatens to penetrate to Black's
rear. Hence a more accurate reply is 27 . . . ..t g7 28 f3 .J:l e4,
preventing 29 .i.f4? because of 29 . . . .J:l xf4 + ! . If 29 d6 (29 .J:l c2
..te5) 29 . . . f7 30 .J:l d l , then 30 . . . ..td4 is playable.
27 . . . <i;f7
Not 27 . . . .J:l e4? 28 ..tg5!
28 .J:l d3 J..fl 6 29 l:!. a3 a5 30 .J:l b3 ..td8 31 .J:l c3 ..tc7 32 a4
<i;f6 33 fl /4 34 J.. c l !5 35 .J:l c2 .J:lg8 36 .J:l e2 ..te5 37
..tb2! ..td4! 38 ..txd4 cd 39 : e7! d3!
The immediate 39 . . . .J:l c8 is weaker: 40 e2 .J:l xc4 41 d3,
and Black is faced with problems (4 1 . . . .J:l xa4? 42 d6).
The amusing 39 . . . f3 40 .J:l xh7 .J:l g l + 41 xgl d3 fails to 42
.J:l f7 + .
40 </;elf .J:l c8!
One more precise move that is indispensable to Black.
41 <i;d2 .J:l xc4 42 xd3 -1--J.
After the forced 42 . . . .J:l xa4 43 d6 : a t 44 c2 (44 c4 .J:l d l )
44 . . . .J:l a2 + 4 5 <i;c3 : a t , the draw i s unavoidable. We now
return once again to the 1 5th match game.
11
li:::J c6
12
h4! (57)
57
B

This is the whole point; White forces his opponent to go into
an ending, as the threat of h4-h5 is quite unpleasant. A factor of
considerable importance is that with the dark-squared bishops
exchanged, the white king will find a comfortable central post on
e3.
12
13
14
-*. xg7
cd
x g7
cd
14 h5 is hazardous : 14 . . . de 1 5 J:l xc3 l:l d8 1 6 .i.d3 g8!, but

not 1 6 . . . llb4 1 7 hg llxd3 + 1 8 l:l xd3 ( 1 8 e2! is interesting) 1 8
. . . 'iFxd2+ 1 9 l:l xd2 l:l xd2 20 xd2 hg 2 1 e5, with approximate
equality.
14
15
16
xd2
e3
'iFxd2 +
:t d8
-*.d7 (58)
58
w
The only way. Postponing development with 1 6 . . . h6 (aiming

to meet 1 7 h5 with 1 7 . . . g5) could be costly to Black after 1 7
.i.b5 .i.d7 1 8 -*.xc6 .i.xc6 1 9 ll:le5.
17
:t b1
As the answer . . . b7-b6 would be bad for Black on account of

-*.a6, which gains control of the c-file, this transfer of the rook is
very useful. Incidentally, 1 7 ..tb5 would fail to the unexpected 1 7
. . . llxd4 ! 1 8 llxd4 e5! with complete equality.
17
18
:t abS
.i.d3
It was worth considering an immediate 1 8 h5!?, since in the

game Black could now have played 1 8 . . . h6 to frustrate this
plan.

18
t'i:J e7
Kasparov probably rejected 1 8 . . . h6 because of 1 9 g4 .i.e8
(nor can he rid himself of his troubles with 19 . . . f6 20 g5 hg 2 1
hg 22 l:t h6 .ll h8 23 l:t xh8 xh8 24 d5, when White has the better
prospects) 20 g5 h5 21 J: hc 1 , and relying on his powerful centre
White can proceed with a long-term plan for pressure on both
flanks.
-
19
h5
19 J: hc l , recommended by Krasenkov, is answered by 1 9 . . .

J:[ bc8 ! and Black has already solved most of his problems.
19
20
21
hg
J: h2 (59)
f6
hg
59
8
Rather slow. White could have set his opponent more problems
with 21 g4!?. The threat to break up Black's kingside structure is
highly unpleasant. Replies which fail to help are 21
g5 (2 1 . . .
.i.c6 22 .i.c4!) 22 e5! t'i:Jd5 + 23 wd2 l:t h8 24 J:[ xh8! wxh8 25
.l:t h 1 + wg8 26 J: h6, and 21
.ll h8 22 .l:txh8 wxh8 23 g5 f5 24
l'i:Je5 .i.e8 25 ..ta6 b6 26 J: c 1 or 26 f3.
It seems Black would have to play 2 1
e5 22 de ..txg4 23
ef+ wxf6. White would then carry out a similar idea to one
which will arise in the actual game (with the difference that the
black pawn is on b6 and the white rook on h 1 ) : 24 t'i:Jd4 J: d7 25
: h4 : bd8 26 e5 + (26 : xg4 : xd4 27 e5 + wxe5 28 : xd4 J:[ xd4
29 f4+ J:[ xf4 30 J: b5 + t'i:Jd5+ 31 n xd5 + wxd5 32 wxf4 draws)
26 . . . wxe5 27 J: b5 + l'i:Jd5+ 28 l:t xd5+ J:[ xd5, but this plan loses
its force because the c6 square has not been weakened. There is
likewise no danger to Black in 24 e5 +!? wg7 (24 . . . we6 25
. .
. . .
. .

llJg5 + !) 25 llJg5 llJd5 + (stronger than 25 . . . .I:Xd3 + 26 xd3
.tf5 + 27 d2! .txb l 28 : xb 1 ) 26 e4 (26 d2? llJf4 27 : b3
.tf5, and it is Black who wins) 26 . . . llJc3 + ! (after 26 . . . .tf5 +
2 7 ri>f3! .txd3 2 8 : h7 + g8 29 : bh 1 , there i s n o defence
against the mate threats) 27 e3, and everything ends happily as
in a good vaudeville : 27 . . . llJd5 + 28 e4 llJc3+ etc.
b6
21
One can understand Black's wish to avoid the unattractive

endgame that results from 21 . . . : h8 22 : xh8 xh8 23 .ta6!
b6 (23 . . . .tc8 24 .tb5! or 24 .tc4) 24 : c 1 , or 23 . . . .tc6 24
.tc4 ..td7 (24 . . . f5 25 : h 1 + g7 26 llJg5 ri>f6 27 f4 fe 28
: h7 : f8 29 g4, threatening 30 llJxe4+ .txe4 3 1 g5 mate; the
only defence is 29 . . . llJd5 +, but then Black loses his rook after
30 .txd5 ed 3 1 : c7!) 25 g4.
22
g4
Kasparov now plays very ingeniously :

22
23
de
e5!?
.t xg4 (60)
60
w
24
ef+
24 : bh 1 has been recommended as stronger. But let us see
whether it is. 24 . . . : xd3 + ! (inadequate alternatives are 24 . . .
.txf3 25 : h7 + ri>f8 26 : h8 + llJg8 27 .tc4! .txh 1 28 ef 'i>e8
29 .te6, and 24 . . . fe 25 llJxe5 .te6 26 : h7 + ri>f6 27 f4 : bc8
28 ttJf3! .tg8 29 ttJg5! .txh7 30 : xh7 : xd3 + 3 1 'i>xd3 : f8 32
e3 llJc6 33 : c7 llJe7 34 llJ h7 + ) 25 'i>xd3 .txf3 26 : h7 + f8
(61) .
27 : 1 h3 (forced, since 27 ef llJg8! 28 : l h3 .txe4 + ! draws; nor
does 28 : 1 h4 help, in view of 28 . . . g5!, and when the rook moves,

61
w
Black again has 29 . . . -*.xe4 + !) 27 . . . -*.g4 (the attempt to weave

a mating net for the white king, with 27 . . . -*.g2 28 J:l. g3 -*.fl +
29 e3 fe 30 .IH3 + g8 3 1 J:l. xe7 J:l. d8, fails to the simple 32
J:l. xe5) 28 J:l. g3 f5 29 e3 J:l. e8! 30 J:l. h8 + ll'lg8, and Black is out
of danger. It remains for us to consider 28 ef J:t d8 + ! 29 e3 (29
c3 J:l. c8 + ! 30 b2 -*.xh3 3 1 fe + e8 32 J:l. xh3 xe7 33
J:t h7 + e6 34 J:l. xa7 J:t f8, and Black has everything in order) 29
. . . ll'lg8 30 J:l. g3 ll'lxf6 3 1 J:t h8 + e7 32 J:l. xd8 xd8 35 e5 ll'ld5+
34 d4 -*.e6 35 J:l. xg6 d7, and Black's defence cannot be
breached.
So the move suggested by the annotators, 24 J:t bh 1 , fails to
justify their hopes.
xf6
24
25
ll'ld4
25 e 5 + g7 (25 . . . e6?! 26 ll'lg5 + ) is not dangerous for

Black.
.l:t b7?! (62)
25
62
w

This move was probably based on miscalculation in the main
variation. But even after the more precise 25
.l:l. h8 (25 . . . g7
leads to an unattractive minor piece ending after 26 .l:l. bh l .l:l. h8
27 .l:l. xh8 .l:l. xh8 28 .l:l. xh8 c;;, x h8 29 c;;, f4) 26 l:t bh l .l:l. xh2 27 .l:l. xh2
c;;, g 7, White can play, for example, 28 f3 (28 f4 is premature
in view of 28 . . . l:t d8 29 c;;, e s g5! 30 lDe6+ ..txe6 3 1 c;;, xe6 .l:l. xd3
32 c;;, x e7 .l:l. d2!), and it is not easy for Black to save himself.
Has Black no other suitable defence, then? Zaitsev suggests 25
l:t e8!?, with the idea of answering 26 .tb5 (26 c;;, f4 .1:1. bd8) with
26 . . . lDf5 + ! 27 lDxf5 (risky alternatives are 27 c;;, f4 g5 + ! 28
xg4 .l:l. xe4+ , and 27 c;;, d 3 .l:l. ed8! 28 .l:l. b4 .l:l. xd4+ ! 29 .l:l. xd4
.te2 + !) 27 . . . .l:l. xe4+ 28 c;;, x e4 .txf5 + . A complete demon
stration of the geometric co-ordination of the black pieces.

26
f3
At this point, White could have put his opponent in a critical

position by playing 26 .l:l. h4!. How is Black to defend?
The obvious-looking 26
.l:l. bd7 (26 . . . .te6 loses to 27 e5 +)
suffers a fiasco after 27 e 5 + ! xeS (27 . . . g5 28 .l:l. xg4+ ! xg4
29 .l:l. g l + is crushing) 28 .l:l. b5 + lDd5 + 29 .l:l. xd5 + c;;, x d5 (Black
comes out a piece down after 29 . . . .l:l. xd5 30 lDc6 + c;;, f6 3 1 lDxd8
g5 32 .l:l. h8) 30 .l:l. xg4. It is notable that this line was indicated
by Mephisto, the World Microcomputer Champion. True, Black
can still struggle on : 30 . . . .l:l. e8 + !? 3 1 f3 (after 3 1 ..te4+
c4!, the black king slips across to the queenside, creating
counterchances; while after 3 1 d2 .l:l. f7 32 f3, Black is not devoid
of hope although he clearly stands worse), and now 3 1 . . . .;;, c s
32 lDb3 + ! or 3 1 . . . e5 32 lDc6+ allows White to re-group
effectively, so Black must play 3 1 . . . .l:l. f7 + or 31 . . . .l:l. f8 + ; though
in either case, after 32 g3, the two pieces are much stronger
than the rook and pawn.
Another try is 26
lDfS + . If now 27 lDxfS gf 28 f4 (if 28 ef,
then 28 . . . .l:l. xd3 + , while after 28 .l:l. g1 .l:l. bd7 29 .tb5 .l:l. e7 30 f3
c;;, g 5 3 1 .l:l. hh 1 Black is guaranteed a draw with 3 1 . . . .l:l. xe4+ ! 32
fe f4+ 33 f2 .l:l. d2 + etc.) 28 . . . fe 29 .txe4 ..tf5!, there follows
a lengthy combination with a happy end : 30 .l:l. h6 + ! c;;, g 7 3 1
..txf5 xh6 3 2 .l:l. h 1 + c;;, g 7 33 .l:l. h7 + f6 3 4 .l:l. xb7 xf5 3 5
.l:l. xa7 .l:l. e8 + ! 3 6 f3 .l:l. e 1 3 7 : f7 + c;;, e6 3 8 .l:l. b7 c;;, fs 3 9 .l:l. xb6
.l:l. fl + 40 e3 .l:l. f3 + ! 4 1 c;;, d4 (capturing the rook leads to a
classic stalemate) 4 1 . . . .l:l. a3 42 .l:l. b2 c;;, x f4 43 c;;, c 5 : as 44 .l:l. b4+
e5 45 a4 .l:l. c8 + , and the draw is elementary.
.

Nonetheless there is a way for White to win. He should answer
26 . . . lDfS + with 27 ef l:l.e7+ 28 .i.e4! gf 29 l:l. xg4 etc.
26
l:l. bd7
If 26 . . . .i.d7 or 26 . . . .i.c8, then 27 f4 is strong.
27 l:l. b4
.t e6
The best defence. 27 . . . .txf3? is refuted by 28 l:l.f2, and 27 . . .
lDc6 by 28 lDxc6 l:l. xd3+ 29 f4 .thS 30 eS + !. If 27 . . . l:l. e8,
then 28 l:l. a4. The matter is more complicated after 27 . .th5
28 f4! lDc6 29 e S + g7 (29 . . . f7 30 .i.c4+ is bad for Black)
30 lDe6 + g8 3 1 lDxd8 lDxb4 32 .i.c4+ lDdS + 33 e4 l:l. xd8
34 .txdS + <i;g7 3S l:l. c2 l:l. d7 36 e6 and 37 eS. At move 30, if
30 . . . <i;h8, then 3 1 .i.xg6, or if 30 . . . <i;h7, then 3 1 l:l. xhS + .
One final possibility i s 3 0 . . . h6 3 1 l:l. b3 l:l. e8 3 2 .txg6! <i;xg6
33 <J;e4 lDxeS 34 l:l. g3 + lDg4 (34 . . . .tg4 3S fS + <J;f6 36 l:l. h6+
<J;e7 37 </;xeS) 3S fS + h6 36 l:l. xhS + <i;xhS 37 l:l. h3 mate.
28 l:l. c2
White could try playing for the win with 28 f4, but Black has
one satisfactory reply : 28 . . . .tg8! (if 28 . . . aS?, then 29 l:l. h7? is
a mistake on account of 29 . . . lDfS + 30 ef ab, but 29 eS + ! f7
30 l:l. h7 + g8 3 1 l:l. xe7! l:l. xe7 32 l:l. xb6, or 3 1 . . . ab 32 l:l. xe6
l:l. xd4 33 l:l. xg6 +, gives White a clear plus) 29 eS+ g7 30 .i.e4,
and the chances are about level.
28 .tbS! looks a much stronger try. There follows 28 . . . l:l. c7
(63)

63
w
29 lDxc6 l:l. c 3 + 30 f4 xe6 3 1 .tc4+ d6 (3 1 . . . f6 is

dangerous : 32 l:l. h7 gS+ 33 g4! and the black king comes under
a strong attack) 32 l:l. d2 + c7 (more precise than 32 . . . <i;cS 33
l:l. xd8 xb4 34 .te6! gS + 3S xgS l:l. xf3 36 l:l. d7 lDc6 37 l:l. c7,

and Black's troubles continue) 33 l:lxd8 cRxd8 34 ..te6 g5 + (34
. . . .!Llc6 can be met by 35 J:l c4! J:l xc4 36 ..txc4 e7 37 g5,
when White's chances are better) 35 cRxg5 J:l xf3 36 J:l d4+ e8
37 J:l d7 (also 37 ..tg4 J:l g3 38 f4 J:l a3 39 ..th5 + cRf8 only
leads to equality) 37 . . . J:l g3 + (37 . . . .!Llc6 is unconvincing in view
of 38 J:l c7 .!Lld4 39 ..td7 +) 38 f6 (an equal position results from
38 cRf4 J:l g6 39 ..tf5 .!Llxf5 40 ef xd7 41 fg e7 42 e5 f8!
43 d5 g7 44 cRc6 xg6 45 a4 cRf6 46 cRb7 aS) 38 . . . J:l g6+
39 cRe5 J:l g5 + (Black loses after 39 . . . .!Llc6+ 40 d5 .!Llb4+ 41
cRd6 .!Lld3 42 J:l xa7, and has a difficult ending after 40 . . . .!Lle7+
41 cRd6 .!Llc8 + 42 cRe5) 40 cRd6 .!Llc8 + 41 cRc7 J:l c5 + 42 b7
J:l e5 43 ..th3 .!Lle7 44 cRxa7 J:l xe4 45 cRxb6 J:l e3, with a clear
draw.
28
aS
29 J:l a4
gS !
30
J:l d6
..tbS
31
32
33
..tel
J:l ac4
..td7
J:l e8
J:l b2
.!LidS+
t-t
Black now has a tiny pull, but neither side can seriously count
on winning.
Game No. 1 5
Karpov-Kasparov
World Championshp (1 7th game)

Lyon 1990
I have said that in my last match with Kasparov the Griinfeld
Defence was played in four games. The first three of them, though
containing fascinating struggles, ended in draws. We have covered
all three together in the context of Game 14. In the seventeenth
match game, which concluded our dispute in this opening, I finally
succeeded in winning. That is why this example merits a separate
number in the book.
1 d4 .!Llf6 2 c4 g6 3 .!Llc3 d5 4 cd .!Llxd5 S e4 .!Llxc3 6 be ..tg7 7
..te3 c5 8 11rd2 0-0
9 n
..t
1
!U p to here, everything coincides with the fifteenth match game.
From the notes to that game, the reader will derive a large amount
of information about this opening line. In that game Kasparov

played 9 . . "ifa5, and encountered distinct problems - as we saw.
Consequently, this time he prefers a different method.
10
ll:lg5!? (64)
.
64
8
Perhaps White can hardly to without this thrust if he is to count

on an advantage. But then I dare say the debate about this opening
line will be continued further, and the assessment will undergo
many alterations.
10
11
cd
cd
ll:lc6
If 1 1 . . . h6, then 1 2 h3 is obligatory.

12
h3
..td7
An inferior choice is 1 2 . . . ..txd4 1 3 ..txd4 ifxd4 14 "if xd4

ll:lxd4 15 hg ll:lc3 + 16 ..t>d2 ll:lxa l 17 ..td3.
13
J:r b1
J:[ c8!
The subtle tactical point lies in the highly attractive, though
not complicated, line 14 J:r xb7 ll:lxd4! 1 5 ..txd4 ..txd4 16 "ifxd4
J:[ c 1 + 1 7 d2 (or 1 7 e2 ..tb5 + 1 8 ..t>e3 J:[ e 1 + ) 1 7 . . . J:r d 1 + !
1 8 ..t>xd 1 ..ta4 + , and White loses his queen. Hence he has to
spend some time bolstering his centre.
14
15
ll:lf3
..td3
ll:la5
..te6
The black pieces have securely fastened onto the c4-point. It is

a rule in many Griinfeld variations that the domination of c4,
in the absence of organic pawn weaknesses, guarantees Black
counterplay.
16 0-0
..t c4
17
J:[ fd1
b5?! (65)

65
w
This might seem to be quite in order; Black strengthens his

hold on c4. But in so doing, he commits a major positional error.
From now on, his queenside pawn chain becomes vulnerable and
causes him a great deal of worry. The modest 1 7 . . . b6! would
have been more appropriate.
18
..tg5
The first encouraging sign. The threat is not only 1 9 ..txc4
.!Dxc4 20 "it"b4 .!Dd6 2 1 e5, but also, in some variations ..txe7,
deflecting the black queen.
18
19
a6
.Z: bc 1
I believe 1 9 : de l !? was also worth considering. Then, for

example, after 1 9 . . . ..txd3 20 : xeS 1Wxc8 2 1 "it"xd3 "it"b7, White
would immediately have the highly effective break 22 a4, and if
22 . . . b4, then 23 ..td2, winning material.
19
20
21
..txd3
"it" xc8
.l:t xc8
'ilfxd3
.l: e8?!
"it"b7 22 "it"a3 .!Dc4 23 "it"xe7 1Wxe7 24
A passive move. 21
..txe7 J:l e8 was more flexible. It is possible that I would then have
had to revert to thy plan of .l:l d 1 -b 1 and a2-a4. After 22 .l: b1,
the reply 22 . . . h6 (l s bad on account of 23 ..td2!. The alternative
22 .z:t cl is less convincing, as after 22 . . . h6 23 ..tf4 tll c4, neither
of White's options promises him anything to speak of: 24 tll d 2
.l:t c8! 25 d5 .!Db2, or 24 d5 J:l c8 25 .!Dd4 tll b2 26 l:t xc8 + 1Wxc8 27
W'd2 .!Dc4 28 .!Dc6 tll x d2 29 tll xe7 + f8 30 .!Dxc8 .!Dxe4.
Another interesting variation is 21 . . tll c4 22 J.. xe7 .l: e8 (22
. . . .!Db2 23 W'b3 tll xd 1 24 J.. xf8 1Wc1 25 .ta3 'ilf a 1 26 "it" c2!) 23
. .

..ta3 tt.J xa3 24 'ilf xa3 l:txe4 25 d5 ..tf8 26 d6 'iii' d 7 27 tt.Jg5, with
complications.
22
l:t c l
'iii' b7
23
24
d5
tt.Jd2
tt.Jc4
A key factor in White's overall strategy. Evicting the last piece

from c4, he seizes the vital file.
tt.Jxd2
24
It was worth considering 24 . . . h6 25 ..tf4 e5 (25 . . . g5 26 ..tg3

tt.Jxd2 27 l:l. c7 'iii' b6 28 'iii' x d2 favours White, as does 27 . . . l:l. c8 28
l:l. xb7 l:l. c l + 29 h2 tt.Jfl + 20 'ilfxfl l:l. xfl 3 1 l:l. a7; another
unsatisfactory line is 25 . . . tt.Jxd2 26 'ilfxd2 g5 27 .:t c7 'iii' b6 28
l:t c6 'ilfd4 29 W'xd4 ..txd4 30 ..td2 .:t a8 31 ..tb4 f8 32 d6! ed
33 l:l. xd6) 26 ..te3 (avoiding the sly trap 26 de?! l:l. xe6 27 tt.J xc4
'ilfxe4 ! with instant equalisation) 26 . . . tt.Jxe3 27 'ilf xe3 l:l. c8 28 tt.Jb3
..tf8 29 tt.Ja5 l:l. xc l + 30 'ilfxc l 'iii' b6 31 tt.Jc6 f6, and Black can
resist stubbornly.
: cs (66)
25
..txd2
66
w
26
l:l. c6!
In view of the variation 26

l:t xc6 27 de 'ilf xc6 28 W'd8 + ..tf8
29 ..th6, Black's position at once becomes critical. Of course, 27
. W'c7 28 'iii' d 7 ..te5 29 ..th6! would also lead to a simple win
for White : 29 . . . 'ilfxd7 30 cd ..tc7 3 1 e5! aS 32 fl b4 33 e2
a4 34 d3 ..td8 35 c4, etc.
. . .
. .
26
27
..te5
..t c3
..tb8

In the event of 27 . . . ..txc3 28 'ii' x c3 .l:!. xc6 29 de iilc7 30 g3 a5
3 1 'ii' c 5 b4 32 iild5, or 30 . . . f6 31 'ii' b3 + f8 32 'ii' e6, the white
queen penetrates decisively to d7.
28
29
'ii' d4
..ta5
f6
Even now White has to play carefully; not 29 .1:1. b6?? .l:!. xc3! 30
.l:!. xb7 .l:!. c 1 + and mates.
29
30
31
'ii' c3
a3
..td6
.l:!. e8
Why hurry? The fruit will ripen of its own accord.

31
32
33
34
35
36
37
g3
'ii' c5
..tc7
..tf4
.l:!. c7
d6!
g7
..te5
h5
.tal
'ii' d7
'ii' d8
The death agony of the black pieces, suffocating on the edge of

the board, now commences.
37
38
39
40
d7
..td2
.l:!. b7
l-0
g5
.l:t8
..te5
On 40 . . . h4, Zaitsev gives 41 ..ta5! 'ii' xa5 42 "it' xe7 + g6 (42

. . . .l:!. f7 43 'ii' xf7 +) 43 "it'h7 + ! xh7 44 d8('ii' ) + . A paradox - to
win Black's queen, ite must first sacrifice his own!
Russian System
Game No. 1 6
Ehlvest-Ernst
Tallinn 1989
1
2
3
4
5
6
7
d4
ll'lf3
c4
ll'lc3
iib3
"ifxc4
e4
ltlf6
g6
J.. g7
d5
de
0-0
J.. g4
The queen move to b3 (or the rarely played "ifa4+) constitutes

what numerous theoretical manuals call the Russian System. But
it incorporates various subsidiary systems which depend on Black's
seventh move. At present, the Ragozin (or Prins) Variation, 7 . . .
ltla6, enj oys the greatest popularity. This is evidently due to its
appearances in the 1 986 and 1987 World Championship matches,
and the subsequent rapid development of its theory. The Smyslov
Variation, 7 . . . J.. g4, also occurred in two of the Karpov
Kasparov match games I shall come to these presently. Among
the rarer and somewhat passive continuations, I would menion 7
. c6 (The Boleslavsky Variation), and 7 . . . a6 with the idea of . . .
b7-b5 (the Hungarian Variation).
-
. .
J.. e3
The most flexible move. The continuations 8 ll'lg5 and 8 ltle5

cause Black no harm.
8
ltlfd7 (6 7)
The critical position of the Smyslov Variation. White's most
popular moves are 9 J: d 1 , 9 J.. e2, 9 0-0-0, and the one he plays
in this game.
9
1Wb3
Russian System 109

67
w
The most active continuation. Instead of 9 b3 White also has

an interesting idea of placing his queen on c5: 9 l:t dl il'lc6 10 .tel
il'lb6, and now 1 1 'ii' c5 'ii' d6 (68) .
68
w
A large number of moves have been seen here : 1 2 "ifxd6, 1 2

il'lb5, 1 2 d 5 , 1 2 h 3 , 1 2 0-0. I would remind you o f the famous
game Botvinnik-Fischer, Varna 1 962, which went 12 h3 ..txf3 1 3
gf .ll fd8 1 4 d 5 il'le5 1 5 il'lb5 'ii' f6 1 6 f4 il'led7 1 7 e5, and now Fischer
gave his opponent an unpleasant shock with 1 7 . "ifxf4 ! . After
some errors . by both sides, however, this fascinating encounter
was eventually drawn.
But White proves to have one other interesting move at his
disposal. Without having been tried out before, it occurred twice
in the return match for the World Championship in 1 986, and for
that reason I have decided to discuss it :
12 e5
A paradoxical decision at first sight. White not only leaves his
pawn on e5 hopelessly weak, he allows a queen exchange too. But
it is not as simple as all that.
. .
110 Russian System

12 . . . "it'xe5 13 de 'D eB
The more natural-looking 1 3 . . . 'Dd7 is met by 14 h3! ..txf3
1 5 gf!, and the 'doomed' pawn on e5 is immune - since if either
knight takes it, 16 f4! wins a piece. Otherwise White will protect
the e-pawn with its neighbour, gaining a clear advantage.
14 h3!
This was played in the seventeenth match game (Karpov
Kasparov, Leningrad 1 986). In the fifteenth game, where the
novelty 12 e5 was used for the first time, I continued instead with
14 'Db5, and after 14 . . . l:t b8! 1 5 'Dxc7 e6! Black managed to
obtain fully equal chances. The point is that the threat to surround
the knight with 16 . . . a6 compels White to lose a tempo with 1 6
'Db5, whereupon the black knight o n c8 quickly transfers itself to
a convenient post. The game concluded : 16 'Db5 ttJ 8e7 17 J:. d2
b6 1 8 cb ab 1 9 ..tg5 'Df5 20 b3 h6 2 1 ..tf6 ..txf3 22 ..txf3 'Dxe5
(only now does Black recover the pawn) 23 ..txe5 ..txe5 24 0-0
J:!. fd8 25 .!:tfd 1 J:. xd2 26 J:. xd2 :t c8 27 g3 :t e l + 28 <;t;>g2 <;t;>f8
29 ..te4 <;t;>e7 -!---!-.
We now return to the seventeenth game :
14 . . ..txj3
If 1 4 . . . ..te6, the sortie 1 5 'Dg5 is unpleasant : 1 5 . . . 'Dxe5 1 6
'Dxe6 fe 1 7 f4.
15 ..txf3 ..txe5
After 1 5 . . . 'Dxe5 1 6 ..txb7 .:. b8 1 7 c6 'Dc4 1 8 J: d7 'Dxb2,
White can choose between two attractive possibilities : 19 'Dd5
and 1 9 'Db5.
16 ..txe6! be 17 ..td4 .i.f4 18 0-0
It is worth considering 18 'De2. A correspondence game Riim
mele-Brummer ( 1 988) continued 18 . . . e5 19 ..tc3 f6 20 l:!. d7 J:. f7
2 1 l:t d8 + <;t;>g7 (White also has a clear plus after 2 1 . . . J:. f8 22
J:. xf8 + <;t;>xf8 23 'Dxf4 ef 24 ..txf6) 22 'Dxf4 ef 23 <;i;>e2, with a
won position. In Rummel's view Black can play more precisely
with 18 . . . ..th6 19 ..tc3 l:. b8 20 J:t d7 ..tg7 21 ..txg7 (2 1 'Dd4 !
i s interesting) 2 1 . . . <;t;>xg7 2 2 b 3 : b 5 23 0-0 .:. xc5 2 4 .:. c 1 .:. x c 1 +
25 'Dxc1 'Dd6 26 .l hc7, with a minimal edge for White.
18 . . . a5?
The correct line is 1 8 . . . e5 1 9 ..te3 ..txe3 20 fe 'De7, as
demonstrated in the game Karpov-Timman, Tilburg 1 986, which
continued :21 .l: d7 'Df5 22 J:. xc7 J:. fc8! 23 l:t d7 l:t d8 24 l:t fd 1 J:. xd7
25 : xd7 'Dxe3 26 l:t c7 .:. b8 27 b3 : d8! 28 'De4 .:. d4 29 'Df6 +
.
Russian System 1 1 1
rt;g7 3 0 J:l. xc6 J:l. d2 3 1 g4 lDc2 3 2 fl lDd4 3 3 J:l. a6 lDf3 !- } .
1 9 'IJ.fe1 a4 20 'IJ. e4 J.h6 2 1 J.e5 a3 22 b3 lDa7 23 J:l. d7 J.cl
24 J:t xc7 J.b2 25 lDa4!
The material balance is restored, but White's threats are
mounting up.
25 . . . lDb5 26 J:l. xc6 J:l.fd8 27 'IJ. b6! J:l. d5
Cunningly thought out; if 28 lD xb2, then 28 . . . J:l. xe5! 29 J:l. xe5
ab 30 J:l. e 1 lDc3, and the b-pawn unexpectedly brings victory to
Black. But avoiding the trap is quite simple.
28 J.g3 lDc3 29 lDxc3 J.xc3 30 c6 J.d4 31 J:l. b7 1-0
Black resigned in view of the inevitable c6-c7.
lD b6
9
Alternatives are 9 . . . c5 and 9 . . . J. xf3
lDc6
10 J:l. d1
Practice has also seen 1 0 . . . e6, and 1 0 . . . e5 1 1 de lD8d7 with
sharp play. With 10 . . . lDc6, Black provokes the advance d4-d5.
.
11
dS
lDeS
12 J.el
lDxf3+
13 gf
J.hS
As practice has shown, 1 3 . . . J.h3 is less convincing in view of
14 J:l. g 1 ! and Black's bishop will soon have to retreat.
,
14
a4
14 lDb5 and 14 f4 are not dangerous for Black. We shall say a

word about 1 4 .l: g l presently.
14
'ii' d7 (69)
In such situations Black usually plays . . . 'ifc8 to defend his b
pawn, but here this is not necessary, since after 1 5 aS lDc8 1 6
'ifxb7 lDd6 1 7 'if c6 'ifh3! Black obtains good counterplay.
69
w
1 12 Russian System
15
J:. gl !
A n interesting idea. I t used to be thought that White should

not allow the queen onto h3. Hence the rook manoeuvre to g 1
was carried out a move earlier; then after 1 4 J:. g 1 'it'd7 (or 1 4 . . .
'it'c8), White could reply 1 5 J:. g3. In the present game, on the
contrary, White 'tempts' the black queen to h3 and then even
further - to h2. The result is mind-bending complications, which
in fact were what attracted me to this game.
Formerly, the normal move from diagram 69 was 1 5 h4, for
example : 1 5 . . . a5 1 6 lt:lb5 lt:lc8 1 7 .td4 (Uhlmann-Kozma,
Zinnowitz 1 967, went 1 7 f4 .txe2 1 8 xe2, and now 18 . . . lt:ld6!
would have equalised) 17 . . . .txd4! 18 lt:lxd4 lt:lb6 1 9 .tb5 'it'd6,
and the chances are level. I should mention that the exchange on
d4 was suggested by Euwe. Long ago, in a game Smyslov
Botvinnik from the return World Championship Match of 1 958,
Black played instead 17 . . . lt:ld6, and White gained a distinct
advantage.
15
Black could also g o after a different pawn - the one o n a 4 but after 1 5 . . . .txc3 + 1 6 be 'it'xa4 1 7 'it'xa4 lt:lxa4 1 8 d2,
White has quite enough initiative for it.
16
17
f4
d2!
'it' xh2
White's intention becomes clear : having lured the enemy queen

into his camp, he aims to exploit its awkward position in order
to work up a kingside attack.
17
.txe2
The sharp variation 1 7 . . . .txc3+ 1 8 be .txe2 19 xe2 'it'h5+

20 d3 J:. ab8 2 1 : b 1 lt:ld7 22 'it'dl is also possible.
l8
lt:lxe2
c6
Black has to play energetically. If 18 . . . 'it'h3, then 19 f5 with

unpleasant threats.
19
20
a5
lt:lg3
lt:ld7
White's plan of surrounding the black queen is quite consistent.

After 20 'it'xb7 lt:lf6! Black would have a promising position.
20
cd
Black sacrifices a piece, but obtains a large number of pawns

in return.
21
J:!. h1
'it' g2
Russian System 113

22
..te2!
de
The only move, in view of the threatened 23 J:t dg l .

23
24
J:t xd7
f5!
e5!
'iff3+
Ehlvest recommends 24 . . . gf at once. Play could continue : 25

.i.. c 5 (it must have been to prevent this move that Black gave
check) 25 . . . f4 26 .i.. xf8 J:t xf8 27 J:t fl (27 J:t hd l e3) 27 . . . .i.. h 6!
(of course not 27 . . . fg 28 "ilfxg3, with a won ending for White) 28
J:t h l .i.. g 7, and White cannot improve the placing of his pieces.
25
..tel
gf (70)
70
w
26
J:t g l !
Once again the rook occupies the g-file, this time with even
greater effect.
26
f4
If 26 . . . J:t ad8 (26 . . . J:t ac8 27 "ili d l !), then 27 Wd5 ll xd7 28
"if xd7 f4 29 lZ:lf5 fe 30 fe is decisive.
27
28
29
30
31
lZ:lf5
'ifxe3!
"if h6
ll xg6
'if h4!
fe
'if xf5
'ifg6
hg
Now there is nominal material equality, but Black's pieces are

very passively placed and he doesn't succeed in co-ordinating
them.
31
32
33
34
a6!
fe
b3
b6
e3
e4
b5
1 14 Russian System
34 . . . -*.c3 + 35 e2 -*.b4 doesn't help : 36 1Wxe4 -*.c5 37
1Wd5!
35 ""el -*.c3 36 1We7 -*.a5 37 .l:l. xa7 .l:l. ae8 38 1Wa3 -*.c3 39
1Wc5 b4 40 .l:l. e7 .l:l. c8 41 1Wd5 1 -0
Game No. 1 7
M. Gurevich-Kasparov
Moscow 1988
1
2
3
4
5
6
7
d4
c4
lt.:lc3
lt.:lf3
1Wb3
1Wxc4
e4
lt.:lf6
g6
d5
-*.g7
de
0-0
lt.:la6 (71)
71
w
Introduced into practice by Ragozin as long ago as the 1930s,

this variation has occurred three times in the encounters between
Kasparov and myself. The nineteenth game of the return match
in 1 986 ended in a beautiful win for White, but in Seville (games
1 5 and 2 1 ) Kasparov succeeded in obtaining equality, and both
games were drawn. I do not know if its occurrence in our matches
was the reason, but since then the popularity of this variation has
noticeably increased, and a good many games have been played
enriching its theory. In the notes to the present game we shall
examine the most interesting practical examples, including all
three of the World Championship games to which I have referred.
In the diagram position, Gurevich developed his bishop on e2.
Strange though it may seem, no fewer than ten other continuations
Russian System 1 15
have also been seen! The moves 8 'ifa4, 8 b4, 8 ..te3, 8 h3, 8 e5,
8 l: b 1 and 8 ..td3 have completely gone out of use. 8 'it'b3, 8
..tg5 and 8 ..tf4 are also employed only rarely, and in any case
they often amount merely to a transposition - the bishop is
brought to e2 a little later. I shall now given one example of each
of these three alternatives that I have said to be the most common.
G. Georgadze-Tukmakov, Odessa 1 989, went 8 'it'b3 c5 9 d5
'it'b6! (9 . . . e6 is the usual continuation. The queen exchange gives
Black a comfortable game) 10 'ifxb6 (perhaps White shouldn't be
in a hurry to exchange; 10 ..tc4 or 10 e5 should be tested) 10 . . .
ab 1 1 ..tc4 e6 1 2 de ..txe6 1 3 ..txe6 fe 1 4 0-0 lt:lb4 1 5 h3 lt:ld7
1 6 l: d l lt:lc2 17 l: b l lt:ld4 18 lt:le1 (better 18 lt:lb5 e5 19 a3, with
unclear play. Now Black seizes the initiative and quickly achieves
a won position) 1 8 . . . lt:le5 19 a4 : ad8 20 ..tg5 l: d7 21 ..th4 h6
22 b3 g5 23 ..tg3 lt:lec6! 24 f1 l: fd8 25 lt:le2 lt:lxe2 26 xe2
lt:ld4+ 27 fl b5! 28 ab lt:lxb5 29 : xd7 : xd7 30 e5 lt:lc3 3 1 : b2
b5! 32 : c2 b4 33 : c l l: d2 34 : a t c4 35 lt:lf3 l: d8 0- 1 .
Flear-Ftacnik, Belgrade 1 989, went 8 ..tgS h6 9 ..th4 c 5 1 0
d5 b 5 ! 1 1 lt:lxb5 (ECO gives 1 1 'ifxb5 : b 8 1 2 'ife2 l:l xb2! 1 3 11Vxb2
lt:lxe4 14 : c 1 'ifa5, with advantage; if now 15 ..txe7 : e8 16 d6
lt:lxd6 17 'it'd2 : xe7 + 1 8 ..te2, Black has the very strong 1 8 . . .
..txc3 19 'ifxc3 'it'b5! 20 lt:l g 1 lt:lb4 2 1 fl : xe2! 2 1 lt:lxe2 ..ta6,
and although White is two exchanges up, Black's attack is
irresistible) 1 1 . . . 'if a 5 + 12 lt:ld2 l: b8 (theory recommends 12 . . .
lt:lxe4, but the rook move is more precise) 1 3 0-0-0 (after the
familiar 13 ..tg3 lt:lxe4 14 'ifxe4 : xb5 1 5 ..txb5 'ifxb5 16 'ife2
'ifxb2 1 7 0-0 lt:lb4, Black has excellent compensation for the
minimal material deficit. But in castling queenside, the English
grandmaster comes under a strong attack) 1 3 . . . lt:lg4! 14 1Wb3
(14 ..tg3 ..td7!) 1 4 . . . c4 15 ..txc4 ..td7 1 6 a4 : res 17 b 1
lt:l c 5 1 8 11V a 3 1Wxa4 ! 1 9 b 3 ..txb5! 0-1 . After 2 0 b a ..txc4 + 2 1
c 1 ..tb2+ 2 2 1W xb2 lt:ld3 +, White suffers big material losses.
Eingorn-H. Olafsson, Reykjavik 1 990, saw 8 ..tf4 c5 9 de 11Va5
(9 . . . ..te6 is also frequently played) 10 e5 lt:ld7 1 1 a3 1Wxc5 (72) .
This position was recently the object of a minor theoretical debate,
associated with three games.
12 1We4 (in Eingorn-Gavrikov, Tallinn 1 989, White played 1 2
lt:ld5, and after the inferior reply 1 2 . . . : e8 he quickly reached a
won position : 1 3 l: d 1 h6 1 4 h3 g5 1 5 1W xc5! lt:ldxc5 1 6 ..te3
etc. Soon afterwards, in Piket-Ivanchuk, Tilburg 1 989, Black
1 16 Russian System
72
w
innovated with 1 2 . . . ttlb6 and seized the initiative : 1 3 1i'xc5 ttlxc5

14 ttlxe7 + h8 15 J:t c l ttle6! 16 .te3 .td7 17 .te2 : ae8! 1 8
.i.xb6 a b 1 9 ttld5 .i.c6 20 l: d 1 f6! with clearly the better po sition.
This time, it is Eingorn who himself changes the complexion of
the game) 12 . . . ttlb6 13 .i.e3 1i'c7 14 : c 1 .tf5 1 5 1i' h4 : ac8 1 6
.i.xa6 b a 1 7 0-0 1i' b1 1 8 ttld4 ttlc4 1 9 .th6 .txh6 2 0 1i'xh6
ttlxe S 21 l: fe1 f6 22 b4 .l:l. fd8 23 ttlxf5 gf 24 l: xe5 .l:l. xc3 25 .l:l. c5
(but not 25 .l:l. xf5 1i'e4 26 .l:l. c5 1i'c2!) 25 . . . .l:l. xa3 (the chances here
are level) 26 .l:l. xf5 .l:l. ad3 27 .l:l. g5 + fg 28 "it'xg5 + f8 29 11t'f5+
e8 30 "it'h5+ f8 31 11t'f5 + -! - ! .
8
.tel
c5
9
d5
9 e5 is met by 9 . . . ttlg4 1 0 h3 cd 1 1 hg de 1 2 be "it' aS, with a

good game for Black.
White gains nothing from 9 de .te6 10 1i'b5 l: c8! 1 1 1i'xb7
ttlxc5 12 1i'xa7 l: c7 13 1i'a3 (if 1 3 1i'a5 ttlcxe4, White can't play
14 ttlxe4 because of 14 . . . : xc1 + !) 1 3 . . . ttld3 + 14 .txd3 1i'xd3,
with a very strong initiative for Black. This whole variation is
given in the opening monographs. In M. Gurevich-Gavrikov,
Moscow 1 990, White varied with 1 1 c6 l: xc6, and then 12 1i'xb7,
but after 1 2 . . . l: xc3! 13 1i'xa6 ttlxe4 14 0-0 l: c7 1 5 .tf4 ttld6!
1 6 .te5 1i'b8 17 .txg7 xg7 18 b3 1i' b4 19 l: ad 1 ttlf5 20 l: fe 1
l: b8 2 1 .tfl l: b6, the game was level.
e6
9
10
0-0
The bishop' sortie 10 .tg5 (73) should also be mentioned.
After 10
ed (stronger than 10 . . . h6 1 1 .txf6 1i'xf6 1 2 e5
.
1i'd8 13 d6!, or 10 . . . 1i' b6 1 1 .txf6 .txf6 12 e5 .tg7 1 3 0-0,
73
B
when practice has shown White's chances t o be better) 1 1 ltlxd5,

Black has two options : 1 1 . . . ..te6 and 1 1 . . . '1Va5 + .
Belyavsky-Kasparov, Belfort 1 988, went 1 1 . ..te6 1 2 0-0-0
..txd5 1 3 .:t xd5 '1Vb6 14 ..txf6 '1Vxf6 1 5 e5 '1Vf5 ! (in Flear
Korchnoi, Lugano 1 988, after 1 5 . . . '1Ve7 1 6 .:t hd 1 : adS 1 7 : xd8
.l:l. xd8 18 .:t xd8+ xd8 19 e6 White obtained the better
chances, although his initiative was gradually neutralised) 1 6 ..td3
11t'c8! 17 .:t d 1 (allowing his opponent to work up a dangerous
attack. The right move was 1 7 .:t d6, with complex play) 17 . . . b5!
18 '1Vh4 ltlb4 1 9 ..txg6 fg 20 .:t d7 '1Ve8! 21 : e7 ..th6 + ! 22 b1
:Z. d8 23 l:. d6 '1Vc6! 24 a3 .:t xd6 25 ed '1Vxd6 26 ab cb 27 '1Ve4 b3
0- 1 .
Farago-Dorfman, Budapest 1 988, continued instead 1 1
11t'a5+ 1 2 ..td2 "it'd8 1 3 ltlxf6+ ..txf6 1 4 e5 ..tg7 1 5 ..tc3 (Gutman
recommends 15 0-0 ..te6 16 '1V c 1 ! '1Vb6 1 7 ..th6, with slightly the
better game) 1 5 . . . ..te6 16 11Vb5 '1Ve7 1 7 0-0 ltlc7 1 8 '1V a4 ..td7
19 "it'c2 ..tc6, with equal chances.
. .
. . .
10
11
ed
eel
: es (74)
The move 1 1 . . . ..tf5 might well transpose after 1 2 ..tf4 J:. e8;
alternative ideas after 1 1 . . . ..tf5 are discussed in Games No. 1 8
and 1 9.
After 1 1
b5 1 2 '1Vxb5! .:t b8 1 3 '1Va4 .:t b4 14 '1Vd 1 ! Black has
no compensation for the pawn.
1 1 . b6 is interesting -- the knight on a6 is at once freed
from its unwanted duty of defending the c-pawn. Tukmakov
Chiburdanidze, Biel 1 988, went 12 .:t d 1 ( 1 2 ..tg5 h6!) 1 2 . . . ltlb4
1 3 a3 ( 1 3 '1Vb3 ..tf5!) 1 3 . . . ..ta6 1 4 '1V b3 ..txe2 1 5 ltlxe2 ltlbxd5
16 ltlf4 c4 17 11t'xc4 '1Vc7 1 8 '1Vxc7 ltlxc7, and the position was
. . .
118 Russian System

completely equal. In Naumkin-Henkin, Moscow 1 989, White
instead played 1 2 ll:le5, but again achieved nothing : 1 2 . . . ll:lb4 !
1 3 ll:lc6 ll:lxc6 1 4 de .t.e6 1 5 1W a4 1Wd4 1 6 1W xd4 cd 1 7 ll:l b 5 ll:ld5!,
and after some sharp play the endgame concluded peacefully.
In Vladimirov-Fette, Groningen 1 989, White answered 1 1 . . .
b6 with the valuable novelty 12 1Wh4!. After just three moves,
Black was in trouble : 1 2 . . . ll:lb4 1 3 .t.g5 .t.f5 ( 1 3 . . . h6 14 .t.xh6
.t.xh6 1 5 1W xh6 ll:lbxd5 16 l:t ad 1 1We7 17 l:t xd5! is also bad for
Black) 1 4 l:t ad l . Vladimirov gives another variation which likewise
favours White : 1 2 . . . ll:lxd5 1 3 ll:lxd5 (or 1 3 .t.g5!) 1 3 . . . 1W xd5
1 4 ll:lg5 h6 15 .t.3 hg 16 .t.xg5.
74
w
12 .t.f4
The fashionable development of the bishop on e3 (though with
Black having played 1 1 . . . .t.f5 instead of 1 1 . . . .l:l. e8) is covered
in Game No. 1 9.
After 12 l:tdl (to prepare a quick advance of the d-pawn) 12
.t.5, we reach a position that occurred twice in the Seville match
(75) . This is the appropriate moment to recall those games.
. .
75
w
The 1 5th game o f the 1 987 match continued :
13 d6 h6
Not 1 3 . . . llJe4? 14 d7!
14 h3
In the 2 1 st match game, I played 14 .i.f4 (see below). A
suggestion of Gutman's is also interesting : 14 a3! llJd7 ( 1 4 . . . llJe4
1 5 .i.e3!) 1 5 1W a2 ! llJb6 1 6 .i.e3! .i.e6 1 7 1W b l . The queen has
escaped from pursuit, and the d-pawn continues to fetter the
opponent's pieces.
14 . . . llJb4 15 .i./4 llJ d7 16 : d2 a6 1 7 1Wb3 b5 18 1W d1 c4 19
a4! llJc5 20 ab llJbd3 21 .i.xd3 llJxd3 22 : xd3! cd
The bishop on f5 has to be left where it is, to restrain the
d-pawn; a weaker choice is 22 . . . .i.xd3 23 : xa6 : xa6 24 ba
1Wa5 25 1Wa4! 1Wxa4 26 llJxa4.
23 llJd5! ab! 24 llJe7+ <i;h7 25 : xa8 1Wxa8 26 llJxf5 gf 27
1Wxd3 'ile4 28 1Wxb5 : a8 29 .i.d2 : dB 30 1W c5 1W e6 31
.i./4 .i.xb2 32 llJh4 .i.f6 33 1W xf5 + 1W xf5 34 llJxf5 h5! 35 g4
hg 36 hg g6 3 7 g2 .i.b2 38 llJe7+ <i;f6 39 llJ c6 : d7 40 llJb8
: dB 41 d7 <i;e6 42 <i;f3 .i.a3 :f--J.
From diagram 75, the 2 1 st game of the Seville match went :
13 d6 h6 14 .i./4 llJ d7!
More accurate than 1 4 . . . llJhS 1 5 .i.e3, when the white bishop
has enticed the enemy knight onto the edge of the board and itself
settled in a good position.
15 : d2
Or 1 5 1W b3 llJb4 1 6 : d2, transposing; alternatively 1 6 .i.c4
1Wf6! 1 7 : d2.
15 . . llJb4 16 1W b3 .i.e6
Better than 1 6 . . . a6 1 7 a3 llJc6 1 8 llJdS!.
17 .i.c4 llJb6 18 .i.xe6 : xe6 (76)
.
76
w
120 Russian System

19 a3?
An unfortunate move. 19 tLlb5 is also bad : 19 . . . .l:. e4 ! 20 ..te3
tLlc4 and Black has an excellent position. Instead, 19
..tg3 tLld3 20 tLlb5 (or 20 tLld5 c4 21 "it"b5 tLlxd5 22 "it"xd5 tLlxb2
23 Wxb7 c3 24 J:l. c2) 20 . . . c4 2 1 "it" a3 leads to double-edged
play. In a simultaneous display with clocks, against six of the
strongest American j uniors (New York 1 988), Kasparov reached
the diagram position again in his game against Rao. But in answer
to 1 9 ..tg3, he didn't j ump to d3 with his knight; he chose the
modest 1 9 . . . Wd7, and after 20 a3 tLlc6 2 1 Wb5?! .l:. c8! 22 .l:. ad 1 ?
(22 tLl d 5 i s stronger) 2 2 . . . ..txc3 23 b e tLle5! 2 4 Wxd7 tLlxf3 + 25
gf tLlxd7, Black achieved a won endgame.
19 . . . tLld3!
The appearance of the knight at d3 causes a certain amount of
confusion in the white camp. Capturing it, obviously, loses to the
fork 20 . . . c4 !
20 ..tg3 c4 21 Wc2 .l:. c8 22 .l:. ad1 Wd7
Not 22 . . . tLlxb2, which fails to 23 Wxb2 tLla4 24 tLlxa4 ! ..txb2
25 J:l. xb2 W a5 26 .l:. b4.
23 h4 f5
Perhaps 23 . . . .l:. c6 or 23 . . . l:t c5 was more accurate. The
exchange sacrifice on d3 would then be ineffective; White would
have to play 24 "it" b 1 and then 25 tLl e l .
24 .l:. xd3 c d 2 5 "it"xd3 tLlc4 26 "it" d5 tLlb6
Evidently, Kasparov's weakened king position is not to his
liking, and he decides not to risk anything. And yet with 26 . . .
<it>h7! Black would have retained the better chances. You may
ask what difference it makes whether the king goes to h8 or h7.
In fact the difference is highly significant. After 26 . . . c;t>h8 (26 . . .
tLlxb2 27 .l:. e 1 .l:. ce8 28 .l:. xe6 .l:. xe6 - 28 . . . "if xe6? 29 d7! - 29
tLlb5 <it>h7 30 tLle5! ..txe5 3 1 ..txe5 tLlc4 32 f4! a6 33 tLld4, with
advantage) 27 tLlb5 tLlxb2 28 .l:. b 1 tLla4 29 tLlc7 .l:. xc7 30 de "it" xd5
31 c8(W)+ c;t>h7 32 .l:. xb7, it is White who has the winning chances.
With the king on h7, the c-pawn would queen without check and
the whole variation would fail.
27 Wd3
More precise than 27 Wb3 Wf7! 28 tLld5 (28 <it>h2 g4) 28 . . .
J:l. d8, with the better chances for Black.
27 . tLlc4 28 "if d5 tLlb6 i-f.
.
Russian System 121

.i.f5
12
Again 12 . . . b6 is possible. In Annageldiev-Arbakov, Uzhgorod
1 988, there followed 13 d6 ( 1 3 a3 li:Jc7 14 d6 li:Je6 gives White
nothing, but 1 3 .:. fd 1 li:Jb4 14 Wb3 .i.f5 15 .:. ac 1 is worth trying)
13
li:J b4 14 li:Jg5 W d7 15 'ifb3 (77) .

77
8
The position looks unpleasant for Black; after 1 5

.i. a6 1 6
.i.xa6 li:Jxa6, his knight has been thrown back again. But the
Moscow master found a powerful resource: 1 5
.:. xe2! (Black
gives up the exchange but activates his bishops) 1 6 li:J xe2 .i.a6 1 7
li:Jc3 (White takes the correct decision t o return the exchange. 1 7
.:. fe 1 i s risky: 1 7 . . . li:J d 3 1 8 .:. ed 1 h6! 1 9 li:Jf3 li:Jh5! with a fine
game for Black, for instance 20 Wd5 .:. e8! 21 .:. xd3 .:. xe2 22 li:Je5
.i.xe5 23 .i.xe5 .:. xe5! 24 'ifxe5 .i.xd3; but not 17 . . . li:Jh5 1 8
li:Jg3! li:Jxf4 1 9 .:. e7 c4 20 iff3, and Black comes t o grief) 1 7 . . .
.i.xfl 1 8 .:. xfl .:. e8 1 9 h3 h6 20 li:Jf3 iff5. A complex position has
arisen, with chances for both sides.

13
.:. ad 1
li:Je4
1 3 . . . iVb6 has been played; White continues 14 ifb5 or 1 4

li:Jh4, with mutual chances.
1 3 . . . li:Jd7 appears to be weaker. Ivanchuk-Kotronias, Lvov
1 988, continued 14 Wb3 li:Jb4 1 5 .:. d2 li:Jb6 16 .i.b5 .i.d7 1 7 .i.g5
ifc8 18 .:. c 1 a5 19 li:Ja4 c4 20 'it'd l li:Jxa4 2 1 'it'xa4 .i.xb5 22 'it'xb5
'it'f5 23 h4 c3 24 be .:. ec8 25 c4, and Black came away empty
handed.
14
..td3
Before proceeding further with the main game, we must recall

the 1 9th game of the return World Championship match ( 1 986),
in which I employed a prepared variation here.
14 li:Jb5! (78)
122 Russian System

78
B
The basic strategic conflict revolves round the d-pawn. If White

succeeds in utilising its potential energy, the initiative will be his.
If Black manages to blockade it securely, his position will be the
more promising. In advancing my knight to b5 I was prepared to
part with the b-pawn, aiming to develop an attack in the centre
after 1 4 . . . J.xb2 1 5 d6 J.f6 1 6 J.d3! and 1 7 ll fe l . Kasparov
refuses the gift.
14 . . fll/6!?
Taimanov recommended 1 4 . . . g5, but after 1 5 J.c1 g4 1 6 e 1
White has a clear positional advantage.
15 J.d3
Considering what happened later in the game, many people
gained the impression that Black's opening set-up had been refuted
outright. But in chess such things rarely happen, and soon after
the match Kasparov himself was to go back to this variation. In
fact, in Belyavsky-Kasparov, Moscow 1 987, it was White who
first departed from precedent by advancing his d-pawn at once.
The ensuing fierce struggle brought about a rapid depletion of
forces : 1 5 d6 J.d7 1 6 g3 g5 1 7 J.e3 h6 18 c7 xc7 1 9 de J.c6
20 d2 ll ac8 2 1 xe4 ll xe4 22 f/J xc5 ll xc7 23 ll d6 b6 24 ll xf6
be 25 ll xc6 ll xc6 26 J.f3 ll xe3 1-1 .
.
15 . . . b4
Not wishing to be condemned to passive defence, Black plays

va banque, but the exchange sacrifice fails to j ustify itself. It may
have been worth taking the pawn with 15
f/J xbl, even though
this involves a definite risk. On the other hand, Black has a way
of sacrificing a pawn himself. This improvement was worked out
by Kasparov, who used it in a clock simultaneous display at a
. . .
Russian System 123

Young Pioneers' tournament. His opponent in this game was the
only master in the contest :
Dzhandzhgava-Kasparov, Baku (Simul) 1 987 : 1 5
J:t ad8!
(possibly it was because of this move that Belyavsky refrained
from 1 5 .id3 and played 1 5 d6) 1 6 J:t de 1 ( 1 6 J:t fe 1 'ifxb2 1 7 J.xe4
J:t xe4 1 8 J:t xe4 .ixe4 1 9 'ifxe4 'if xb5 20 d6 deserved consideration.
Also worth noting is 16 llJxa7 'ifxb2 17 llJb5 llJb4 18 J.b1, when
1 8 . . . llJxd5 19 J:t xd5 J:t xd5 20 'ifxd5 'ifxb5 appears to fail after
2 1 llJe5! However, 2 1 . . . J.xe5 22 .ixe5 llJd2! is better for Black
ed.) 1 6 . . . 'ifxb2 1 7 llJc7 llJxc7 1 8 J.xc7 llJd2! (a pretty resource,
eliminating all inconveniences. In the sharp endgame, Kasparov
easily outplays his young opponent) 19 J:t xe8 + J:t xe8 20 llJxd2
'if xd2 21 J.xf5 gf 22 g3 J.d4 23 d6 J:t e 1 24 g2 J:t xfl 25
'ifxfl 'ifxa2 26 'ifb5 g7 27 h3 'ifd5 28 'ife8 'ifc4 ! 29
g2 W"c2 30 W" e 1 W"a4! 31 W" d2 1Wc6+ 32 f3 J.f6 0-1 .
At move 1 5, Black also has the interesting 1 5
J.d7. This
occurred in lvanchuk-Dorfman, Lvov 1988. The further course
of the game was fascinating : 1 6 J.e5 J.xb5 1 7 W" xb5 J:t xe5 1 8
llJxe5 llJd6! 1 9 llJg4 'iff4 2 0 1W d 7 c4 2 1 g 3 llJc5 2 2 W" c 7 W" xg4 2 3
'ifxc5 'if d 7 2 4 J.e2 .ixb2 2 5 1W b4 c 3 26 J.d3 a5 27 W"f4 J.a3
28 J:t fe 1 J.c5 29 g2 b5 20 h4 h5 31 'iff6 'ifd8 32 'ifxc3 J.b4
33 1We5 J.xe 1 34 J:t xe l . The concentrated tactical crossfire has at
last culminated peacefully; the players agreed a draw a few moves
later.
16 llJc7 llJxd3 1 7 llJxe8 J:t xe8 18 'ifxd3!
To Black's misfortune, 18 . . . llJxf2 or 18 . . . llJg3 would fail to
the counter-stroke 19 'ifb5.
18 . . . 'if xb2
For the exchange Black has a pawn and an active position, but
the passed pawn on d5 has yet to reveal its full potential.
19 J:t de1
The straightforward 1 9 d6 J:t d8! 20 'ife3 h5 21 J:t b 1 W"xa2 22
J:t xb7 W"d5 23 J:t xa7 llJxd6 24 .ixd6 W" xd6 25 llJg5 is unpleasant
for Black and was undoubtedly worth playing. But I decided to
pin the knight, restricting the activity of the enemy pieces. White
incidentally threatens 20 g4. Of course, 19 J:t fe 1 doesn't work in
view of 19 . . . W" xf2+ 20 h 1 "ifxe 1 + ! 2 1 J:t xe l llJf2 + 22 g1
J:t xe1 + and 23 . . . .ixd3.
19 . 'ifb4
The decisive mistake. After 1 9 . . . llJf6 20 J:t xe8 + llJ xe8 2 1 'ife3,

. . .
. .
124 Russian System

White can realise his material advantage without trouble, but with
1 9 . . . 'lfxa2 20 11V b5 (20 g4 c4!) 20 . . . l: d8 2 1 W xb7 11Vxd5, Black
can hold out. In his notes to the game, Kasparov pointed out that
after 22 11V xd5 l: xd5 23 g4, Black has the neat 23 . . . ltlf6!.
20 ltld2!
Playing to exploit the pin is the chief factor in White's strategy.
20 . . . '1V a4 21 W c4
Forcing the queens off, whereas 2 1 ltlxe4 is not so clear : 2 1 . . .
l: xe4 22 l: xe4 .txe4 23 11Vd2 c4, with counterplay.
21 . . . '1Vxc4 22 ltlxc4 .tc3
Black would lose at once with 22 . . . b5 23 ltld2 ltlf6 24 l: xe8+
ltlxe8 25 l:e1 etc.
23 ltl d2 .txd2 24 .txd2 .td7 (79)
79
w
To many onlookers during the game, Black seemed to have

'wriggled out', and indeed done more than that. The bishop on
d2 is attacked, and Black threatens . . . ..tb5. But White is able to
give the exchange back . . . .
25 J..f4! .tb5 26 f3! g5
After 26 . . . .txfl 27 xfl ltlf6 28 l: xe8 + ltlxe8 29 .te5 !, the
d-pawn finally shows its full capability : 29 . . . f6 30 d6! and Black
has to give up a piece.
27 .txg5 .txf1
Nor is 27 . . . ltlxg5 any better : 28 l:[ xe8 + .txe8 29 h4 !, winning
the knight.
28 x/1 ltld6 29 J.. e 7 ltlc8 30 .txc5 l: d8 31 J: e5 f6 32 l:/5
b6 33 J.. d4 ltle7 34 .txf6 J: xd5 35 J: g5+ J: xg5 36 J.. xg5 ltlc6
3 7 e2 !7 38 d3 e6
After 38 . . . ltlb4 + 39 c4 ltlxa2 40 b3, the black knight is
trapped.
Russian System 125

39 c4 ttJe5 + 40 d4 ttJc6 + 41 c4
Here the game was adjourned, and Black resigned without
resuming. After 41 . . . ttJe5 + , White can achieve his ends with
either 42 b5 or 42 d4 ttJc6+ 43 e4, and the white pawns
are not to be stopped.
We now return to the main game, which concluded very quickly :
14
15
16
17
be
1 hb5
1ha6
..txc3
b5
ttJxc3
The correct plan, involving an exchange sacrifice, was demonstrated

by Gurevich in another game : 1 7 'it' c4!? ttJxd 1 1 8 J:l xd 1 ..txd3 19
J:l xd3 'it'b6 20 J:l b3 'it'f6 2 1 g3 ! J:l ad8 22 ..tg5 1Wd6 23 ..tf4 'it'f6
24 a3!; M. Gurevich-Kotronias, Reykjavik 1 988. The black knight is
badly placed, and the activity of White's pieces assures him of
more than enough for the exchange.
17
18
..txd3
'it'xd3
If 1 8 J:l xd3, then 1 8 . . . ttJe2 + 1 9 h 1 ttJxf4 is good for Black.

18
ttJe2 +
ttJxf4
19
h1
'it'd6 (80)
20
'it'c4
In an earlier game Ptin-Lau, played in a postal tournament
( 1 975), Black continued with 20 . . . 'it'f6, leading to equality.
It is strange that thirteen years later (this game was played in
the 55th USSR Championship), Grandmaster Gurevich - who is
one of Kasparov's seconds, too - should bring about this position
again. It is only now that Black plays a new move, and the
resulting ppsition can already be assessed as good for him.
80
w
126 Russian System

21
.l:l. fel
After 21 g3 (2 1 lt)g5 'ife5! 22 lt)f3 'iff5) 21 . . . lt)h5 22 .l: fe l (22

lt)d2 lt)f6 23 lt)b3 .l:l. ac8) 22 . . . lt)f6 23 .l: xe8 + .l: xe8 24 .l: c l .l: c8
25 .l: d l .l: d8, Black has a slight advantage.
Apparently, 21 lt)d2 lt)xd5 22 lt)e4 (but not 22 lt)b3 .l: ad8) 22
. . . 'iff4 (22 . . . 'ife5) 23 'if xc5 .l: xe4 24 'ifxd5 .l: ae8 would have
maintained equality (Kasparov). Incidentally, after 2 1 lt)d2 a bad
line for Black is 21 . . . .l: ad8 22 lt)e4 'ife5 23 f3 g7 24 d6 'ifb2
25 :t g l 'ifb4 26 'if c l ! lt)e2 27 'if a l + lt)d4 28 .l: b l ! 'ifa5 29 : be l
9 b6 3 0 .l: gd l .l:l. e6 3 1 lt)xc5 .l: exd6 32 .l: c4, and White wins;
Annageldiev-Chudinovskikh, Yalta 1 989.
White's unfortunate rook move leads to a quick catastrophe.
21
22
: xe1 +
lt)xe1
White has also a difficult position after 22 l:t xe1 lt)xd5 23 .l: d 1
lt)b6 2 4 9xf7 + xf7 2 5 .l:l. xd6 e7, but now Black succeeds in
increasing his advantage decisively.
22
23
.l: b8!
a3
23 lt)c2 loses to 23 . . . 9xd5!. If 23 g3, Black has 23 . . . : b4!,

while 23 lt)d3 is met by 23 . . . xd3 24 9 xd3 : b2.
23 . . . : b2 24 f3 9e5! 25 'ife4 9g5 26 g3 9h5 27 h4 lt)e2 28
9e8+ g7 29 d6 lt)xg3 + 30 g1 lt)e2+ 31 fl 9f5 32 9 xe2
9 h3+ 0-1
Game No. 1 8
Belyavsky-Tukmakov
Odessa 1989
1 d4 lt)f6 2 c4 g6 3 lt)cJ d5 4 lt)f3 J.. g7 5 9 b3 de 6 9 xc4 0-0
7 e4 lt)a6 8 J.. e2 c5 9 d5 e6 10 0-0 ed
11
ed
J.. f5
12
J.. f4 (81)
The more 'modest' bishop development with 1 2 J.. e 3, which

has been seen frequently of late, is examined in Game No. 1 9.
A standard position in this variation. The contemporary line
12
.l: e8 was examined in detail in Game No. 17 (where the
move-order was 1 1 . . . l:t e8 1 2 J.. f4 J.. f5). Before demonstrating
the valuable novelty employed by Tukmakov in the present game,
let us recall the 'classical' continuation 12 . . . 'if b6, which in our
day has been undergoing further development. An entertaining
. .
Russian System 127

81
B
instance of this occurred in the 1 987 USSR Championship in

Minsk, where Grandmaster Gavrikov, one of the most noted
specialists on the black side of the Griinfeld, employed this queen
move three times.
Belyavsky-Gavrikov continued 13 h3? (an unsuccessful inno
vation) 1 3
'if xb2! 14 g4 i,c2! 1 5 .l:t ac 1 d7 1 6 b5 i.a4! 1 7
d6 (Black also has the advantage after 1 7 .l:t b 1 b6 1 8 .l:t xb2
xc4 19 .1:1. bb 1 , but now the White position becomes highly
critical) 17 . . . b6 1 8 'ife4 ( 1 8 'ifd3 b4) 1 8
.l:t ae8 etc.
A more logical plan was tried in Gurevich-Gavrikov from the
same event : 13 i.e5.
This manoeuvre is the most dangerous to Black. 1 3 a4 can
be met by 1 3 . . . 'if b4 or 1 3 . . . 'ifa5, and 1 3 .l:t ad 1 by 1 3 . . . 'ifxb2.
13 . . .l:t ad8 (82)
. . .
. . .
82
w
Alternatives are 13
.l:t fe8 and 13
e8, but taking the b
pawn is risky; after 13
'if x b2 14 e4 'if b6 15 d6, or IS i.xf6
.i.xf6 1 6 .l:t ab l , Black is in great danger. On the other hand, in
Ivanchuk-Lputian, Irkutsk 1 986, White played the mistaken I S
. . .
128 Russian System

J: ab1? t'Llb4 1 6 ..txf6, and now with 16 . . . ..txe4 ! 1 7 J.. x g7 ( 1 7
J.. e 7 J.. x d5!) 1 7 . . . J.. xb 1 1 8 J.. xf8 J.. x a2 1 9 'it'xc5 J: xf8, Black
could have achieved a winning position.
14 J: fd1
Perhaps 14 J: ad 1 was more exact.
14 . . . J: fe8
Against Lputian (this is the last in the trio of examples),
Gavrikov played 14 . . . t'Lle8 1 5 t'Lle4 'it'a5 1 6 J.. x g7 <l;xg7 1 7
J: ac 1 t'Lld6 1 8 'it'b3 (it was better t o play a t once 1 8 'it'c3+ 'it'xc3
19 t'Llxc3, with equality) 18 . . . J.. d 7 19 'it' c3 + 'it'xc3 20 t'Llxc3, and
now after the retreat of the knight to b8 or c7 Black would not
be threatened with anything. Instead he continued with 20 . . .
J: fe8?, when 2 1 .i.xa6 enabled White to obtain a big endgame
advantage.
15 'it'h4
The immediate 1 5 b3 is more accurate.
15 . . . t'Lld7 16 J.. xg7 <l;xg7 1 7 b3 'it'a5 18 J: acl t'Llf6
Black has no problem whatsoever, and after 19 ..t b5 .i.d7 20
..txa6 'it' xa6 21 t'Lld2 h6 22 'it'c4 'it'a5 23 fl 'it' b4 24 t'Llde4 'it'xc4
25 be t'Llxe4 26 t'Llxe4 b6 27 f2 f5 28 t'Llc3 f4 29 g3 the players
agreed a draw.
After these three games by Gavrikov, the impression was formed
that in the 1 2 . . . 'it'b6 line White has no particular prospects of
gaining the initiative. An attempt to refute this opinion was
undertaken in the game Vladimirov-Popovic, Moscow 1 989,
which made a substantial contribution to theory. From diagram
82, play went : 14 d6! 'it'b4 (Black goes into a sharp endgame; 14
. . . 'it'xb2 loses to 15 t'Lld5) 15 J: ad 1 'it'xc4 1 6 J.. xc4 t'Llb4 (this
game has been thoroughly annotated by the winner and by
Grandmaster Makarichev and International Master N. Andri
anov. I shall quote some of the interesting variations they give.
At this point, Black could perhaps have got off with a slight scare
by continuing 1 6 . . . t'Lle4 1 7 J.. x g7 <l;xg7 1 8 t'Llxe4 J.. xe4 19 J: fe 1
J.. xf3 20 gf) 1 7 h 3 ! (not prophylaxis, but preparation for an attack)
1 7 . . . a6 (it is now too late for 1 7 . . . t'Lle4, in view of 1 8 g4! t'Llxc3
19 be J.. e4 20 .i.xg7, and White wins) 1 8 a3 t'Llc6 19 g4 .i.d7 ( 1 9
. . . ..te6 i s worse : 2 0 ..txe6 fe 2 1 J: fe 1 t'Llxg4 2 2 h g J: xf3 2 3 ..txg7
<l;xg7 24 t'Lle4 J: f4 25 t'Llxc5 J: xg4 + 26 fl . with a solid advantage
to White) 20 J: fe l b5 2 1 ..td5! (Black has all the time been
refraining from . . . t'Llc6xe5 in view of the dangerous pressure
Russian System 129

against f7. But his attempt to drive the bishop off its diagonal or
block it in with 2 1 .ta2 c4 does not succeed. White is prepared
to exchange this bishop only if it can be replaced on its active
square by a knight) 2 1 . . . ll:lxe5 (2 1 . . . b4 would be answered by
22 ab cb 23 .txf6 .txf6 24 ll:la4 and 25 ll:lc5, but 2 1 . . . J:t fe8!
was stronger) 22 lLixe5 b4 23 lLixd7! (the point of White's plan lies
in the variation 23 . . . J:t xd7 24 lLi a4 J:t fd8 25 lLixc5 J:t xd6 26 lLib7
ll:lxd5 27 ll:lxd8 J:t xd8 28 .l:l d3 .txb2 29 ab followed by J:t ed 1 ,
winning. However, after 24 . . . lLixd5! 25 J:t xd5 J:t fd8 2 6 ll:lxc5
J:t xd6 27 J:t xd6 J:t xd6 28 J:t e 8 + .tf8 29 ab g7, Black would
have every chance of holding on) 23 . . . ll:lxd7? (now Black has no
counterplay at all) 24 ab cb 25 ll:la4 .tf6 26 J:t e4 a5 27 .tc6 ll:le5
28 .tb5 J:t b8 29 J:t d5 (Black's extra pawn on the queenside has
no significance and White wins the ending, creating quite a good
textbook example in the process.) 29 . . . lLif3 + 30 g2 lLig5 3 1
J:t c4 lLie6 3 2 d 7 g 5 33 fl 'li1g7 3 4 .tc6 .td8 35 J:t f5 b 3 (35 . . .
f6 was more tenacious) 36 ..td5 ll:lf4 37 J:t c8 .tf6 38 J:t xf8 J:t xf8
39 .txb3 J:t d8 40 ll:lc5 J:t b8 41 ll:le4 ..txb2 42 J:t xf7 + g6 43
ll:ld6 1 -0.
Returning to diagram 8 1 , we observe that the Odessa
grandmaster excluded himself from the debate on the variation
12 . . . "ifb6 1 3 ..te5, and instead employed a very valuable novelty,
which, it appears, he had prepared specially for this USSR
Championship.
12
ll:ld7! (83)
83
w
By this means (the rook on f8 and queen on d8 are both left

where they are for now), Tukmakov succeeds in drawing his
opponent into a tactical whirlpool.
130 Russian System

13
d6
Although this advance fits into White's plans, in the present

situation it is rather committal. It was worth considering 1 3 .1g5,
or 1 3 lHd 1 .!l:lb6 14 1i'b3 .!l:l b4 1 5 .J:l d2!
tL!b6
13
14
1i'b3
.!l:lb4
15
.J:l acl
Perhaps the queen should have returned home with 15 1i'dl.
15
.i.e6
Black has many possibilities - 15 . . . a5, 15 . . . .!l:ld3, 15 . . .
.i.d3 - but the move played is not bad either.
16
'ifa3
Again it was safer for the queen to return to its own camp : 1 6
W d 1 .!l:l 6d5 ( 1 6 . . . .i.xc3 1 7 be .!l:lxa2 1 8 .J:l c2 .!l:la4 1 9 Wd2) 1 7
.!l:lxd5 .!l:lxd5 1 8 .i.g3 .i.xb2 1 9 .J:l xc5 .!l:lc3 20 'ifc2 .!l:lxe2 + 2 1
'ifxe2 .i.a3 22 .J:l c3 .i.b4 2 3 .J:l c2, and White's chances are a little
better (Tukmakov).
16
.!l:lc4
.i.xc4
17
.i.xc4
18
.J:l fd1
b6
White's hope of 18 . . . .!l:ld3 1 9 .J:l xd3 .i.xd3 20 'ifxc5 is

disappointed.
19
tL!e4
Here again, White had to play 1 9 'ifa4 a6 20 .J:l d2, and then
'ifa4-d l . It is becoming clear that the white queen is out of the
action.
19
20
.i.d5!
.!l:lfg5
Black also has a big advantage after 20 .!l:lc3 .i.xf3 2 1 gf 1tf6

22 .i.g3 'ifxf3 23 .!l:ld5 'ifxa3 24 ba .!l:lc6! 25 .!l:le7 + .!l:lxe7 26 de
.J:l fe8 27 .J:l d7 .i.d4. Now after 20 . . . h6, White is counting on
bringing his queen across to h3 (2 1 Wh3 .J:l e8! 22 d7 .J:l e7).
Wd7! (84)
20
After this cool rejoinder, White's pieces turn out to be most
precariously placed.
21
.i.d2
Against the piece sacrifice 2 1 .!l:lxc5 be 22 .J:l xc5, Black has the
decisive 22 . . . 'ifb7!, since 23 .J:l cxd5 tL!xd5 24 Wf3 fails to 24 . . .
1i'b4 !
21
.!l:lxa2
Russian System 131

84
w
22
23
24
25
26
ll a 1
ltlc3
..txc3
..txg7
xg2
h6
ltlxc3
..txg2
xg7
hg
As well as two extra pawns Black has the better position, as

the d-pawn presents no threat.
27
1tg3
llfe8
Black could also calmly play 27 . . . g4.

28
29
30
1t xg5
h3
ll a3
.l:!. e4
.l:l. ae8
.l:t 8e5
1tg3
ll f3
ll d5
.1:1. 4
.l:t e8
a5
ll 4e6
30 . . . a5 is simpler.
31
32
33
34
This loses at once - the f7 point is invulnerable, and Black wins

the d-pawn. 34 h4 was more stubborn.
34
35
36
37
1tb3
1Wf3
g1
0- 1
ll d8!
1tc6
.l:!. d7
.l:t exd6
Game No. 1 9
Bareyev-Lputian
Lvov 1990
1 d4 ltlf6 2 c4 g6 3 ltlc3 d5 4 ltlf3 ..tg7 5 1t b3 de 6 1t xc4 0-0
7 e4 ltla6 8 ..te2 c5 9 d5 e6 10 0-0 ed 1 1 ed ..tf5
132 Russian System

12
.i.e3
The alternatives 1 2 .i.f4 and 1 2 l:t d 1 were explained in the

context of Game No. 1 7.
12 .i.g5 is also occasionally seen. An old game Brinck-Claussen
Jakobsen, Denmark 1 970, continued 12 . . . h6 1 3 .i.xf6 11fxf6 1 4
l:t ad 1 l:t ad8 1 5 -*.d3 l:t fe8 1 6 .i.xf5 11fxf5, and White had no
more than a minimal plus.
12
l:t e8 (85)
85
w
13
l:t acl
The alternative is 13 l:t ad1 ; let us look at a few interesting
examples.
A game Ree-de Boer, played some time ago (Amsterdam 1 983),
went 1 3
.!D e4 14 .!Dxe4 l:t xe4 1 5 11fc1 .!Db4 16 a3 .!Dc2 17 .i.g5
f6 18 11fxc2 fg 19 .i.d3 lH4 20 11fxc5 g4 21 11fe3, and White gained
a clear advantage. An improvement is 1 5 . . . 11rb6 1 6 b3 l:t ee8 1 7
.!Dd2 .!Db4 1 8 .!Dc4, with a minimal plus for White.
Farago-Kozul, Montpellier 1 989, went 13
1i' b6 14 b3 l:t xe3 !?
( 1 4 . . . 11t'b4 leads to unclear play. In Ree-Chandler, Helsinki 1 984,
Black ended up in a difficult position after 1 4 . . . .!Dg4 1 5 -*.d2
l:t ad8 1 6 l:t fe 1 .!Db4 1 7 .!Da4 11rd6 1 8 .i.f4 ! 1Wf8 19 .i.c7 l:t d7 20
d6. Black also stands worse after 1 4 . . . .!De4 1 5 .!Dxe4 l:t xe4 1 6
11t'b5) 1 5 fe .!Dg4 1 6 .!Da4 11rd6 ( 1 6 . . . 11fa5 1 7 e4 .!De3 1 8 11rd3 .!Dxd 1
1 9 ef .!Dc3 20 .!Dxc3 is in White's favour) 1 7 11rf4 ! (better than 1 7
e4? -*.d7 1 8 11f c 1 b 5 1 9 .!Db2 .!Db4, with sufficient compensation
for the exchange, for example : 20 .!Dd3 .i.d4+ 21 <ot h 1 .!Dxa2,
with advantage to Black) 17 . . . 11fxf4 1 8 ef .!Db4 ! (but not 18 . . .
.!De3, in view of 1 9 .i.xa6 ba 20 .!Dxc5) 1 9 l:t d2 .!De3 20 l:t c 1 .!Dbxd5

. . .
Russian System 133

2 1 i.d3 lL!xf4 22 i.xf5 lL!xf5 23 lL!xc5 i.h6 24 J: e 1 , and the
sharp play eventually led to a peaceful conclusion.
In another game, the Yugoslav grandmaster Kozul played the
white side, and had to contend with the novelty 13
b6. There
followed 14 h3 J: c8 (better than 14 . . . 1Vb6 1 5 b3 1Vb4 1 6 lL!a4,
or 14 . . . lL!e4 15 lL!xe4 l: xe4 16 1V c 1 , hitting c5 and h6) 15 J: d2
( 1 5 1Vb3! is stronger) 1 5 . . . lL!e4 1 6 lL!xe4 l: xe4 1 7 1V c l c4! 1 8
lL!d4 i.d7 1 9 lL!e6!? fe 20 de J: xe6 2 1 i.xc4 1Ve8 22 J: fd l i.b5?
(22 . . . l: xc4! is correct; Kozul gives the variation 25 1Vxc4 i.b5
24 1Vb3 i.a4 25 "ii' x b7 i.xd 1 26 J: xd 1 J: xe3! 27 fe 1Vxe3 + 28
h1 lL!c5 29 1Vxa7 h7 30 J: fl 1Ve2 31 J: f7 1Ve1 + , with a draw)
23 i.xe6+ 1Vxe6 24 1V xc8 + 1Vxc8 25 ll d8 + 1Vxd8 26 ll xd8 +
f7 27 b3 lL!b4 28 a4 i.c6 29 i.xa7, and White won the ending;
Kozul-Popovic, Yugoslavia 1 989.
Van der Sterren-Miralles, Lyon 1 990, went 13
J: c8 (again
Black innovates - this time successfully) 14 h3 lL!b4 1 5 lL! e 1 lL!e4
1 6 lL!xe4 (better is 1 6 a3 lL!xc3 1 7 be lL!c2 1 8 lL!xc2 i.xc2 1 9 ll d2
i.e4, with unclear play - Miralles) 1 6 . . . J: xe4 1 7 1Vb3 a5! 1 8 g4
i.d7 19 l: c 1 a4 20 1V d 1 a3! 2 1 ba lL!xa2 22 i.d3 J: xe3 23 fe
lL!xc 1 24 1Vxc 1 c4 25 i.c2 (Black also has the advantage after 25
i.e4 1Ve7 26 .i.f3 i.h6 27 lL!c2 c3. Now the c-pawn settles
matters) 25 . . . 1Vg5 26 1Vd2 c3 27 1Vd3 J: c5 28 .i.b3 c2! etc.
13
lL!e4
The immediate 1 3 . . . 1Vb6 deserves attention.

. .
14
1Vb3
Another possibility is 1 4 J: fd 1 lL!d6 ( 1 4 . . . 1Vb6 1 5 lL!a4) 1 5 1Vf4,

with complex play.
14
1Vb6!
15
i.b5
A level game results from 15 J: fd 1 1V xb3 16 ab J: ad8 1 7 lL!xe4
i.xe4 18 i.xa6 ba 19 J: xc5. The chances are approximately
equal after 15 i.xa6 ba 1 6 lL!xe4 i.xe4 1 7 l: xc5 1V xb3 18 ab
i.xb2.
15
16
J: ed8
lL!h4 (86)
Rather a risky move; 1 6 J: fd 1 , defending the central pawn, was

safer.
16
lL!xc3
17
be
i.e4
lL!c7
18
c4
134 Russian System

86
B
19
-*.a4
'lff6!
The awkward position of the knight on the edge of the board

makes itself felt.
20
l0f3
Not 20 f3, on account of 20 . . . -*.xd5!

20
21
22
-*.xf3
gf
-*.d2
b6
22 g2 'lf h4, followed by . . . -*.e5, is bad for White.

l0e6?!
22
This looks pretty, but now White extricates himself from a

tricky situation. His defence would have been more difficult after
22 . . . l0e8! 23 .*.xeS (otherwise 23 . . . l0d6) 23 . . . l:l. xe8 24 l:l. fe l
'lfh4.
23
24
25
26
27
28
de
"ife3!
"ife4
ef+
l:l. cdt
'lfg4!
.I:Xd2
l:l. d6
l:l. b8
xf7
'lfg5+
More exact than 28 hl -*.d4 29 -tc6 g7 30 -*.d5 l:l. f8. A

level endgame now arises.
28
'lf xg4 + 29 fg l:l. bd8 30 l:l. xd6 l:l. xd6 31 l:l. dt l:l. xdl + 32
-txdt g5 33 g2 e6 34 f3 -*.d4 35 e4 -*.xf2 36 h3
.
t-t.
5 4 ..t f4
System
Game No. 20
Lukacs-Ftai!nik
Stara Zagora 1990

The reader will no doubt recall that for the eleventh game of
the return World Championship Match in 1 986, a brilliancy prize
was awarded - to both players. Indeed, that game was one of the
most fascinating played with the Griinfeld in recent years. We
shall, of course, examine it here (in addition to other important
games with the same variation) in the notes to the principal game.
1
2
3
4
d4
c4
ll:lc3
J. f4
ll:lf6
g6
d5
Theory sometimes refers to this variation as the Classical Line.

4
.i g7
5
e3
c5
This system was used in the first game of the 1 986 match, but
with a different move-order (e2-e3 was postponed for three
moves) : 4 ll:lf3 .ig7 5 .if4 c5 6 de "ifa5 7 l:l c 1 (previously, 7 cd
had been played; Kasparov responds to the novelty very precisely)
7 . . . de! 8 e3 "ifxc5 9 'ira4+ ll:lc6 10 .ixc4 0-0 1 1 0-0 .id7 1 2
'irb5 'irxb5 1 3 J.xb5 l:l ac8 1 4 l:l fd 1 l:l fd8 1 5 h 3 h 6 1 6 fl a6
1 7 J.e2 J.e6 1 8 l:l xd8 + l:l xd8 19 ll:le5 ll:lxe5 20 J.xe5 l:l d2
!- ! .
de
Winning a pawn with 6 J.xb8 l:l xb8 7 'ira4+ .id7 8 'ir xa7 is
too risky: 8 . . . cd 9 'ir xd4 0-0 1 0 cd 'ira5 1 1 'ird2 b5! 1 2 .id3 b4
1 3 ll:lce2 'irxd5.
6
7
'ira5
l:l d (87)
136 4 J.f4 System

87
B
Here too, the pursuit of material gains brings trouble : 7 cd

ltlxd5 8 'irxd5 J.xc3 + 9 be 'irxc3 + 1 0 e2 'irxa 1 1 1 J.e5 'ir b 1
1 2 J.xh8 J.e6 1 3 'ir d 3 'irxa2 + , with a very strong initiative for
Black.
Instead, Timman-Kasparov, Belfort 1 988, continued as follows :
7 'ir a4 + 'irxa4 8 lt!xa4 0-0!?
After 8 . . . ltle4 9 J.xb8 .*.d7 10 f3 .i.xa4, the game is level.
But Kasparov is capable of finding new ideas even in seemingly
peaceful positions.
9 lt!f3
Salov-Korchnoi, Brussels 1 988, went 9 .l:l c 1 .i.d7 1 0 ltlc3 de
1 1 J.xc4 lL!a6 1 2 lL!f3 lL!xc5 1 3 e2 .l:l ac8 14 .te5 a6 1 5 a3 lL!a4
1 6 lL!xa4 .txa4 1 7 ltld4 ltle4 1 8 .txg7 xg7 !-! .
lt! e4 10 J.e5 J.d7 1 1 lt!c3 lL!xc3 12 be de 13 .txe4 .l:l e8
9
14 .td4 (88)
.
88
B
An improvement. In an earlier game between the same

opponents (Amsterdam 1988), White played the ineffective 14
4 .i.f4 System 137

.t.d5? .t.c6 15 .t.xc6 : xc6 ( 15 . . . tll x c6! 1 6 .t.xg7 xg7 17
0-0-0 tll d 8 18 ll f8 1 9 ll hd 1 e8 20 ll 7d5 tll e 6 would, in
Kasparov's view, have given Black a plus) 1 6 .t.xg7 xg7 17
: b 1 ll c7 1 8 tll d4 tll a6 19 c6 b6 20 f4 tll b8 2 1 : b4 tll xc6 22 ll c4
ll ac8 23 tll b 5 ll d7 24 tll d4 ll dc7 !- t .
14 . . . e5! 15 .t.xe5 ll xe5 16 .t.xg7 xg7 1 7 .t.b3 ll xc3 18
0-0
White could have kept the game equal with 1 8 d2 or 1 8 ll d l .
Now Black seizes the initiative.
18 . . . tll a6! 19 tll e5 .t.e8 20 .t.d5 ll e7 21 : ab1 tll e5 22 e4
ll d8 23 llfcl ll de8 24 g4 f6 25 tllf3 b6 26 tll d4 J.d7 27 f3 tll d3!
28 ll xe7 ll xe7 29 ll d1 tllf4 30 !2 f8 31 .t.b3 e7 32 tll e2
tll xe2 33 xe2 ll c3
Black has an obvious endgame plus, and he quickly converts it
to a win.
34 h4 h6 35 e5? J.b5+ 36 !2 fe 37 ll d5 ll e5 38 ll xe5 be 39
g5 hg 40 hg J.d3 41 J.g8 .t.f5 42 .t.b3 .t.e6 43 J.e2 .t.xa2 44
.t.xg6 a5 45 e3 a4 0-1.
7
8
9
10
J.xc4
de
0-0
Wxc5
tll f3
.t.b3
10 tll b 5 will be mentioned in the notes to Game No. 2 1 .
10
tll c6
Game 2 1 also introduces the idea of playing the knight to a6.

11
0-0
WaS (89)
89
w
I should explain that the specific game which we are examining

reached the diagram position by way of 7 tll f 3 0-0 8 ll c 1 de 9
.t.xc4 Wxc5 10 .t.b3 WaS 1 1 0-0 tll c6 . For convenience in the
138 4 J.f4 System

layout of material, I have given the opening moves in a different
order.
12
h3
J.f5
12
1W a6!? is also interesting. It was first played by Mecking
against Ribli in Manila 1 976. The idea is familiar - Black tries to
eploit the weakness of the d3-square in the White camp. If 13
ll)d4, Black should not play 13
l:.d8 1 4 J.c7 l:l. d7 1 S J.xf7 +
xf7 1 6 1W b3 + e8 1 7 ll)e6! with the unanswerable threats of
ll)cS and ll)xg7+ . In Farago-Gavrikov, Amsterdam 1 987, Black
replied correctly with 13
J.d7, and after 14 ll) dbS a draw was
agreed; in the event of 14 . . . l:l. ac8, the position js completely
equal.
Another possibility (after 12 . . . 1Wa6) is 13 ll) a4 l:l. d8 14 ltlcS
l:l. xd 1 1 S ll)xa6 l:l. xfl + 16 xfl J.fS 17 ll)cS. In a game Schneider
Itkis, USSR 1 987, White gained a big advantage after 17 . . . b6?
1 8 g4 ! J.c8 1 9 ll)e6!, but a new move 1 7 . . . ll)aS! was played in
Mikhalchishin-Gavrikov, Lvov 1 987. There followed 1 8 J.c2
J.xc2 19 l:l. xc2 ll)dS 20 J.eS ll)b4 21 l:l. c3 b6 22 J.xg7 xg7
23 ll)d3 ll)xd3 24 l:l. xd3 1 - 1 .
Finally, there i s a third option for White : 13 ll)e5 ll)xeS 1 4
J.xeS J.e6 1 S J.xe6 1Wxe6 1 6 J.d4 ( 1 6 J.h2 l:l. ac8 1 7 1Wa4 a6
1 8 iraS ll)e4 is good for Black; Sinkovics-Krasenkov, Budapest
1 988) 1 6 . . . l:l. fc8 17 1Wf3 1W c6 18 1Wxc6 l:l. xc6 19 ll)bS l:l. xc 1 20
l:l. xc 1 a6, with an equal game; Levitt-Gulko, St John 1 988.

. . .
. . .
13
1We2
The ninth game of the 1 986 match went 13 ll) d4 J.d7 14 Wel
ll)xd4 1 5 ed e6 16 J.dl "ii' b6, resulting in equal chances. Another
four moves - 1 7 l:l. fd 1 J.c6 1 8 J.e3 iraS 19 ..td2 Wb6 20 J.e3
WaS - and we agreed a draw.
An attempt at strengthening White's play was made in Peturs
son-Ivanchuk, Reggio Emilia 1 989/90 :
16 J.e5 J.c6 1 7 l:l.fd1 l:l.fd8 18 1We3 l:l. d7 19 1Wg5
It is worth considering the sharp 19 d5 J.xd5 20 J.xf6 J.xf6
2 1 ll)xdS ed 22 Wf3 ..txb2 23 l:l. xdS! Ivanchuk examines two
alternatives for Black: 19
ed 20 Wd4 ll)e8 21 ..txg7 .!L\xg7 22
J.xdS l:l. ad8 23 WeS J.xdS 24 l:l. xdS l:l. xdS 2S .!LlxdS, and 19
.!Llxd5 20 ..txdS J.xdS 21 J.xg7 xg7 22 .!Lle4! l:l. e8 23 Wf4
Wd8 24 .!LieS l:l. de7 (but not 24 . . . eS? 2S 1Wd2 l:l. d6 26 .!L\xb7 J.xb7 27
1Wxd6 irgS 28 f3! J.xf3 29 1W d2, with advantage to White) 2S
ll)e4 l:l. d7 - with equality in both cases.
. . .
. . .
4 j_f4 System 139

19 . . . "ild8 20 liJ a4 h6 21 "flg3 liJh5 22 "fle3 "flh4 23 liJc5 j_xe5
24 de
The correct line was 24 11t' xe5 : e7 25 liJxe6!? j_xg2 26 oi>xg2
fe 27 j_xe6+ oi>h7 28 : c7 liJf4 + 29 oi>fl : ae8 30 d5 liJxh3 3 1
Wg3 W xg3 3 2 : xe7 + : xe7 3 3 fg liJg5, although Black's chances
would still have been slightly better.
24 . . . : xd1 + 25 j_xd1 liJf4 26 g3 1Wxh3 27 j_f3 liJd5
And Black went on to win.
13
14
liJe4
eS (90)
liJd5
In Rizhkov-Epishin, Leningrad 1 986, Black played 1 4 . . . liJc5
at once, which led to equality after 15 j_c4 e6 16 b4 1i'a3 1 7
liJe7 + liJxe7 1 8 be j_e4.
90
w
In the 'main' game which we are examining, the white queen's

bishop calmly retreats to h2 - the familiar theoretical continuation.
In the 'sensational' encounter in London ( 1 9th game, 1 986), I
played the unexpected 1 5 : xc6!? In the digression which now
follows, we shall examine that game as well as the current
theoretical verdict on White's exchange sacrifice.
15 : xc6 ef
So Kasparov decides to decline the sacrifice. Many commentators thought that accepting it would virtually lose outright : 15 . . .
be 16 liJ e7 + h8 17 liJxe5 j_xe5 1 8 liJxc6, but now after 1 8 . . .
"ild2! 19 j_xe5 + f6 it is White who suffers material loss. I was
intending 1 7 liJxc6, and if 17 . . . "ilb6 ( 1 7 . . . Wc5 1 8 liJcxe5 "fle7
19 liJd4), then 18 liJcxeS. After 18 . . . j_e6, White has three choices :
1 9 j_xe6 1i'xe6 20 b3, with two pawns for the exchange; 19 liJc4
140 4 J.f4 System

.i.xc4 20 .t.xc4 lDc5, with unclear play; or 19 11fc2 .t.xb3 20
11fxe4 .i.e6, as in Gavrikov-Kochiev, Tallinn 1 987.
16 l:t c7 .i.e6
If 1 6 . . . fe 1 7 11fxe3 .i.xb2, then 1 8 lDd4 is strong; or if 1 7 . . .
lDd6, White has 1 8 l:t d 1 , with mounting pressure.
1 7 1i'e1!
The natural-seeming 17 : xb7 .i.xd5 1 8 l:t b5 would give Black
the better game after either 18 . . . 11t'a6 19 .i.xd5 l:t ae8 or 1 8 . . .
lDc3 1 9 be .i.xf3 20 gf 11fxc3. In a later game Szihigyi-Schmidt,
Hungary 1 986, Black played the even stronger 17 . . . lDd6!
(depriving the rook of the b5-square) 18 lDe7 + ..th8 19 lDc6 'it'c5
20 .i.xe6 11f xc6, and White came out a piece down.
The queen move to e 1 was planned in my pre-game analysis,
but later it was established that 17 lDe7 + is stronger : 17 . . . ..th8
1 8 l:t fc 1 -*.xb3 1 9 ab, with the initiative ( 1 9 . . . fe 20 11fxe3 lDd6
21 'it'f4).
1 7 . . . 1i' b5!
The endgame after 17 . . . 11fxe1 1 8 l:t xe 1 .i.xb2 19 lDe7 + h8
20 .t.xe6 fe 21 ef l:t xf4 22 l:t xb7 .t.c3 23 l:t e2 is unattractive for
Black.
18 lDe7+ h8 19 -*.xe6 fe 20 11t'b1!
The white queen has made an unusual tour, 1i' d 1 -e2-e 1 -b 1 ,
and now unexpectedly aims a t the opponent's kingside.
20 . . . lDg5!
The only move. 20 . . . 11t'b6 is simply answered by 2 1 : re t , while
if the knight goes anywhere else, 2 1 lDxg6+ is decisive.
21 lD h4! (91)
91
B
After 2 1 lDxg5 (2 1 lDd4 11fe5) 2 1 . . . 11f xg5 22 ef J: xf4 23 l::t xb7

l:t e8 24 lDc6 11fc5 25 lDxa7 .i.d4, Black is dangerously active.
4 j.,f4 System 141

21 . toxh3+ !?
In this extremely sharp duel, each of us was hoping to outwit
the other. No doubt for that reason, Kasparov refrains from 2 1
. . . fe, after which I could have forced a draw with 2 2 tohxg6+ hg
23 toxg6 + g8 24 toe7 + .
22 h2
Of course not 22 gh 'lfg5 + 23 tog2 f3.
22 . . . 'lfh5?
A serious error. 22 . . . toxf2! 23 l:t xf2 fe 24 .l:l xf8 + .l:l xf8 25
h3 e2 26 'lfe4 'lfh5! 27 .l:l c4 g5 28 'lfxe6 gh leads to a draw.
23 toexg6+
If 23 xh3, Black replies 23 . . . g5.
Many observers thought that capturing with the other knight
was stronger : 23 tohxg6 + hg 24 'lfxg6. As we shall see, this
variation is indeed unpleasant for Black, but it happens to be
unimportant which knight takes on g6 first.
23 . . . hg 24 'lfxg6
An inaccuracy throwing away the win! 24 toxg6+ g8 25
toe7 + h8 26 'lf g6! (rejecting the perpetual check with 26 tog6+)
would have given White a clear advantage; this is the same
position that was arrived at in the previous note. I shall now
quote the interesting variations given by Halifman : 26
'lf eS (26
. . . .l:l f5 is met by 27 gh fe 28 'lfxh5 + .l:l xh5 29 tog6 + h7 30
fe!) 27 xh3 fe (27 . . . .l:l f6 28 g4!) 28 'lfg4! .l:l f6 29 'lfh4+ j_h6
30 f4! 'lfxb2 3 1 .l:l b 1 !, etc.; a more stubborn defence is 26
'lfh7,
but here again, after 27 gh .l:l f6 (27 . . . j.e5 28 .l:l c5 ! ; or 27 . . . fe
28 'lf xe6 .l:l f6 29 'lf xe3 with two extra pawns) 28 'lfg4 (28 'lfxh7 +
xh7 29 .l:l g 1 is also good), Black is in a critical position since
taking on e3 is no good: 28 . . . fe 29 fe! .l:l xfl 30 tog6 + g8 3 1
'lf xe6+ .l:l f7 3 2 'lf xf7 mate.
24
'lfe5!
An ingenious retort which gets Black out of danger. White
cannot now play 25 xh3, as the rook on c7 is en prise (whereas
with the knight on e7, the capture 25 . . . 'lfxc7 would be unplayable :
26 'lfh5 + and mates).
25 .l:l/7
Not 25 .l:l xg7 fe+ 26 f4 'lf xg7 27 'lf h 5 + g8 28 gh 'lfxb2 + 29
h1 .l:l f7!, or 26 'lfg3 'lfxg7 27 tog6+ g8 28 toxf8 tog5! 29
tod7 .l:l d8 30 toe5 e2 3 1 .l:l e l .l:l d l 32 tod3 'lfh7 + 33 gl 'lfxd3
34 'lf xg5 + f7 etc. On the other hand, 25 'lfc2 would have forced
.
. .
. .
. . .
142 4 f4 System
a draw : 2S . . . fe+ 26 cot>xh3 'iPg8 27 f4 e2 28 fe ef('if 29 .l: xg7+
cot>xg7 30 'ifg6+ etc.
25 . . . .l: xf7! 26 'ifxj7 fiJg5!
After the game I discovered that the computer had indicated
26 . . . 'if bS here, but there would follow 27 fiJg6 + .,Ph7 28 fiJe7
'ife8 29 11t'xe6 fOgS 30 11t'fS + cot>h6 31 .l: h 1 ! and in spite of his
extra peice Black has no defence.
27 fiJg6+ cot>h7 28 fiJxe5 liJxf7 29 liJxf7 'it>g6 30 fiJd6 fe
The position has become simplified, the tension has abated, and
Black has even emerged with a slight edge, though its significance
is purely symbolic.
31 fiJc4 ef 32 .l: xf2 b5 33 fiJe3 a5 34 .,Pg3 a4 35 .l: c2 .1:/8 36
'it>g4 d4 37 .l: e2 xe3 38 l:t xe3 .1:/2 39 b3 .l: xg2+ 40 .,Pf3
.l: xa2 41 ba J-J.
15 .i.h2
fOeS
15
fiJf6 is very strongly answered by 16 .l: xc6! fiJxdS 17 .l: d6
fiJe7 1 8 e4, with a clear advantage; Pinter-Rogers, Szirak 1 986.
But a more popular continuation is 15
e6 16 .l: fd1 (92) . Let
us look at some interesting games with this line from recent years.

92
B
Farago-Smejkal, Baden-Baden 1 98S, went 16

.,PhS 1 7 c4
.l: ad8 1 8 b4 fiJxb4 1 9 fiJxb4 .l: xd 1 + 20 .l: xd 1 11t'xb4 2 1 xeS,
with a slight advantage to White.
Gleizerov-Vakhidov, Uzhgorod 1 988, went 16
.l: fe8!? 1 7
11t'c4 fiJf6 1 8 e4 .l: ac8 1 9 11t'c3 11t' xc3 20 fiJxf6 + xf6 2 1 .l: xc3, and
a draw was agreed.
Basin-Kozlov, Naberezhnie Chelny 1 988, continued 16
.l: fd8
1 7 11t'c4 (the exchanges with 1 7 11t'e 1 .l: d7 1 8 11t' xaS fiJxaS 1 9 fiJc7
fiJxb3 20 fiJxa8 fiJxc1 21 .l: xc 1 f6 are to Black's liking; Rajna
4 J..f4 System 143

Nikoloff, St John 1 988) 17 . . . lll d4? (the correct line is 1 7 . . . lll f6
1 8 e4 .l:l. ac8 1 9 lll g 5 lll d4 20 lll e7 + f8 2 1 tll xe6+ xe7 22
lll x d8 .l:l. xc4 23 J.. xc4 J.. h6, and Black's chances are no worse;
Huzman-Dorfman, Lvov 1 988) 1 8 ed J.. x d5 1 9 "it'e2 J.. h 6 20
l:t c2 J.. f4 2 1 J.. x d5 "it'xd5 22 lll xe5 J.. x e5 23 J.. x e5 lll g 5 24 "it'g4
lll e6 25 "ifh4 g5 26 "it'h6 "it'e4 27 "it'f6 f8 28 .l:l. cd2 "it'g6 29 d5
lll c 5 30 d6 1 -0.
16
e4!?
An interesting novelty. After 1 6 J.. c4 .:. adS? 1 7 b4! lll xb4 1 8

lll e 7+ h8 1 9 lll xf5 gf 20 lll x e5, White had a big advantage in
Vainerman-Epishin, Norilsk 1 987, but Black has the much better
16 . . . e4!, with chances of seizing the initiative.
16
: adS! (93)
1 6 . . J.. xe4 loses to 1 7 J:t xc5 J.. x f3 1 8 "ife3 J.. h 6 1 9 "it'xh6.
White also has the initiative after 16 . . . J.. e6 17 J.. c4, with
a2-a3 to follow, or 16
lll xb3 1 7 ab J.. e6 1 8 .l:l. a 1 "it'c5 1 9
.l:l. fc 1 "it'd6 2 0 .l:l. d 1 "if b8 2 1 b4. But 16 . . tll xe4 !? leads t o sharp
play.
.
. . .
93
w
17
"it'e3
An inaccuracy. In Ftacnik's view, an improvement is 1 7 .l:l. fd 1

J.. xe4 1 8 .l:l. xc5 J.. x f3 1 9 "it'e3 J.. x d 1 2 0 .l:l. xa5 J.. x b3 2 1 a b lll x a5
22 lll e7 + h8 23 J.. x e5 lll c6 24 J.. x g7 + xg7 25 "it'c3 + f6 26
lll xc6 be 27 "it'xc6, with somewhat the better ending.
17
18
19
20
ab
J:t c5
"it'xc5
lll xb3
J.. xe4
"it' xc5
20 lll f6+ J.. xf6 2 1 "ifxc5 .l:l. d5 22 "ife3 was more exact.
144 4 J.f4 System

20
21
22
23
24
25
'lfe4
gf
J.g3
'lfb5
'lfxb7
: xd5
.i.xf3
: fd8
: dl
l0d4
l0e2 +
Black could have retained the better chances with 25 . . . Jl xb2

26 'lfe7 : f8 27 'lfxa7 l0xf3 + 28 g2 : xb3.
26
27
l8
29
gl
xg3
'lfxa7
l0xg3
J.b6
: xb2
'lf b6
The position has levelled out, and the game gradually heads
towards a draw.
29
30
: at
: d3
g7!
Precisely played. Not 30 . . . e4? 3 1 'lf b8 + J.f8 32 : aS : xf3 +

33 h4. Also 30 . . . J.f4+ 3 1 g2 : dxb3 is dangerous in view
of 32 : aS + g7 33 'lfd8, with an attack (Ftacnik).
31
32
33
34
: a6
'lfe7
gl
J.g5
J.f4+
: dl
'lf b6
l-t
Game No. 2 1
Belyavsky-Gavrikov
Moscow 1988
1 d4 l0f6 2 e4 g6 3 l0e3 d5 4 J.f4 J.g7 5 e3 e5 6 de 'lfa5 7
: ct de
In the fifth game of the return World Championship Match,

London 1 986, Kasparov played 7 . . . l0e4 here. The game was
quite interesting, so let us take this opportunity to give it in full :
7 . . . l0e4 8 cd l0xc3 9 'lf d2 'lfxa2 10 be
This variation came into use after the famous game Petrosian
Fischer, Candidates final 1 97 1 , which continued : 10
'lfa5 1 1
J.e4 l0d7 12 l0e2 l0e5 1 3 J.a2 J.fS 1 4 J.xe5! J.xe5 1 5 l0d4
'lfxc5 16 lOxfS gf 1 7 0-0, with a dangerous attack. Afterwards,
various alternatives were proposed for Black; the most effective
of them belongs to Mikhalchishin : 12
l0xe5 1 3 0-0 0-0 14 f3
. .
4 J.f4 System 145

e5! 1 5 J.g3 b5 1 6 J.a2 1Wb6! 1 7 h1 a5, with adequate
counterplay.
Recently the investigations have switched to 12 10f3, which so
far has brought variable success. In Agzamov-Gulko, Frunze
1985, the interesting continuation was 1 2 . . . /Oxc5 ( 1 2 . . . 0-0 1 3
0-0 /Oxc5 1 4 J.e5 J.xe5 1 5 /Oxe5 f6 1 6 .J:I. a 1 10e4 led t o sharp
play in Razuvayev-Mikhalchishin, Minsk 1 985) 1 3 J.e5 J.xe5!
(a game between the same opponents a few months earlier in
Sochi had ended quickly with 1 3 . . . 0-0 1 4 0-0 f6 1 5 .J:I. a 1 1Wd8
16 J.c7! 1Wd7 17 d6 + e6 18 /Od4 1Wf7 19 .J:I. a5 b6 20 .J:I. xc5! be 21
/Ob3 1Wd7 22 1Wd3! .J:I. d8 23 1We4 1 -0) 1 4 /Oxe5 f6 15 10f3 0-0 1 6
/Od4 /Oe4 1 7 1Wb2 /Od6 1 8 J.a2 J.d7 1 9 0-0 .J:I. ac8 2 0 e4! 11'c5
21 .J:I. fe 1 , with the initiative.
10 . . . 'll xd2+ 11 xd2 10d7 12 J.b5
After 12 c6 be 1 3 de ltlb6 (or 13 . . . ltlf6), White achieves nothing.
12 . . . 0-0 13 J.xd7
This time, 1 3 c6 would be met by 1 3 . . . ltlc5. Now Black obtains
the two bishops, but it soon becomes clear that they will both be
constricted.
13 . . . J.xd7 14 e4 f5
The alternative 1 4 . . . .J:I. ac8 is weaker : 1 5 J.e3 f5 1 6 f3.
15 e5 e6!
A critical moment. The line that had previously been tested was
1 5 . . . .J:I. ac8 16 c6 ( 1 6 e6 J.a4 17 c4 .J:I. xc5 18 J.e3 .J:I. c7 19 /Of3
.J:r. fc8 20 d3 b5! favours Black) 16 . . . be 17 d6 ed 18 ed .J:r. f6,
with the advantage; Schmidt-Gross, Naleczow 1 984. Why does
Kasparov reject it? The answer is that after 1 5 . . . .J:I. ac8 White
has the much stronger 1 6 c4 ! .J:I. xc5 17 J.e3. In Seirawan-Adorjan,
New York 1 987, there followed 17 . . . .J:I. c7 ( 1 7 . . . .J:I. a5 18 f4 e6 1 9
d6) 1 8 10 f 3 b 6 1 9 c 5 b e 2 0 .J:I. xc5 .J:I. xc5 2 1 J.xc5 .J:I. c8 2 2 J.xa7,
. and White won.
16 c4 .J:r.fc8
After 1 6 . . . g5 1 7 J.xg5 .i.xe5 1 8 10 f3 J.g7 1 9 .J:I. b 1 and
.J:I. he 1 , White would complete his development while maintaining
powerful pressure.
1 7 c6!
The extra pawn cannot be retained, but in returning it White
extracts the maximum profit : he creates a passed pawn and limits
the scope of Black's rooks and light-squared bishop.
146 4 j_f4 System

1 7 . . be 18 d6 c5
Now the light-squared bishop acquires a little freedom, but the
dark-squared one begins to suffocate. Perhaps the fate of the bishops
should have been decided 'the other way round', with 1 8 . . . g5 1 9
j.xg5 j.xe5 2 0 c 5 .l:l. cb8, and Black has hopes o f counterplay.
19 h4! h6 20 li:Jh3! (94)
.
94
B
Kasparov must have been examining the more natural-seeming

20 li:Jf3, which he could answer with 20 . . . j_c6!, leading to
complex play with mutual chances. But I arrived at a solution to
the position on mathematical lines. The white knight is heading
for its ideal square, d3, by the only suitable route! Once the knight
reaches its destination, the black bishop on g7 will be interned for
good inside the cage formed by the pawns on d6 and e5 and the
bishop on f4. At the present moment, White forestalls the freeing
advance . . . g6-g5.
20 . . . a5 21 f3 a4 22 .l:l. hel!
Over-protecting the pawn on e5. The hasty 22 li:Jf2 could be
met by 22 . . . g5! 23 hg hg 24 j_h2 f4 25 li:Jd3 .te8 26 j_g l ,
when the situation i s fairly obscure.
22 . . . a3 23 li:Jf2 a2 24 li:Jd3 .l:l. a3 25 .l:l. al g5
The alternative 25 . .1:1. b8 is inadequate on account of 26 : ee l
g 5 2 7 h g h g 2 8 li:Jxc5 (but not 2 8 .txg5 .1:1. bb3 2 9 li:J xc5 .1:1. b2 +)
28 . . . j.a4 29 li:Jxa4 : xa4 30 j_xg5 j.xe5 31 d7.
26 hg hg 27 j_xg5 (95)
Now after 27 . .1:1. b8, the contest could have ended in a study
like draw : 28 j_f4 .l:l. bb3 29 li:Jxc5 .l:l. b2 + 30 c1 l:l xg2 3 1 j.d2
j_h6! 32 j.xh6 .l:l. c3 + 33 d1 .l:l. d3 + ! 34 li:Jxd3 j.a4+ 35
c1 .l:l. c2 + , with perpetual check. But there is quite a simple
.
4 J.f4 System 147

95
B
refutation of the 'study' in 28 e2! l:t bb3 29 .lt.:lxc5 ll b2+ 30

fl . and White wins.
27 . . . f7 28 J..f4 l:l. b8 29 l:l. ecl J.. c6 30 l:l. c3 l:l. a5 31 l:l. c2
l:l. ba8 32 .lt:lcl ]-{)
8 J.xc4
0-0
9
.lt.:lf3
A couple of years ago, when Black was achieving good results

with this Griinfeld variation, the rare move 9 .lt.:le2 was tried out
in a game Barlov-Gulko, New York 1 988. There followed 9
1t' xc5 10 1t'b3 .lt.:lc6 1 1 .lObS 1t' h5 1 2 .lt.:lc7, and now after 1 2 . . .
1t'a5 + 1 3 1t'c3 1t'xc3 + 1 4 .lt.:lxc3 Black equalised.
In lnformator 45, Gulko considers 1l
l:t b8, and judges that
White stands better on the basis of 1 3 J.xf7 + l:t xf7 14 l:t xc6
1t'a5 + 1 5 .lt.:lc3 .lt.:le4 1 6 .lt.:ld5! e6 1 7 l:t c7. But this assessment was
later refuted by Krasenkov in his game against Toth, Mazatlan
1 988 : 16 . . . .lt.:lxc3 1 7 .lt.:lxc3 ( 1 7 l:t xc3 J.xc3 + 1 8 be J.e6 19 J.xb8
J.xd5 20 1t'b2 J.xg2 2 1 l:t g 1 was more tenacious, although even
then Black would be better) 1 7 . . . be 1 8 J.xb8 ( 1 8 1t'xb8 J.xc3 +
1 9 be 1t'xc3 + 20 e2 1t'c4+ 2 1 f3 1t'e6! 22 e2 1t'xa2 + , and
White is in a bad way) 1 8 . . . e6!, and in view of the threatened
19 . . . l:t b7 and 20 . . . l:t xb2, White loses his bishop on b8. A game
Zlochevsky-Krasenkov varied with 1 7 .lt.:lxe7 + f8 1 8 l:t xc3
J.xc3 + 19 be l:t xf4! 20 ef xe7 21 0-0 J.e6 22 l:t e 1 1t'b6, and
White resigned after ten more moves.
A little later, Krasenkov ascertained that all these variations
are redundant, as the simple 1 5 . . . e5 (attacking the bishop and
threatening 16 . . . be 1 7 1t'xb8 1t' xc7) settles matters at once. After
1 6 l:t xf6 J.xf6 1 7 .lt.:ld5 ef 1 8 .lt.:lxf6 + g7, White's position is
completely hopeless.
.
148 4 i./4 System

1i'xcS
i.b3
1 0 ll:\bS looks like an active move, for example : 10
11t' b4+ 1 1
ll:\d2 ll:\a6 ( 1 1 . . . ll:\e4 is more precise) 1 2 a3 11t' a5 1 3 b4 11t'd8 1 4
i.e5 with advantage; Huzman-Tseshkovsky, Tashkent 1 987.
However, Black has the powerful retort 1 0
.i.e6! (96) .
9
10
. . .
96
w
1 1 ll:\c7 (after this, Black seizes the initiative. Unclear play results
from 1 1 .t.xe6 11t'xb5 1 2 .i.c4 11t'xb2 1 3 0-0; if 1 2 .i.b3, then 1 2
. . . ll:\e4) 1 1 . . . .i.xc4 1 2 b3 (in Popchev-Lalic, Bosna 1 988, White
opted for 1 2 ll:\d2, but after 1 2 . . . b5 1 3 b3 ll:\d5 14 ll:\xa8 ll:\c3 1 5
.l:l. xc3 i.xc3 1 6 be 11t'xc4 ! 1 7 11t'e2 .i.xd2 + 1 8 11t'xd2 ll:\c6 1 9 11t'e2
11t'b4+ his position was hopeless) 1 2 . . . 11t'a5 + 1 3 11t'd2 11t' xd2+ 1 4
ll:\xd2 i.d3 1 5 ll:\xa8 ll:\d5 1 6 ll:\c7 .l:l. c8 1 7 ll:\f3 ll:\xf4 1 8 ef i.b2
19 d2 i.xcl + 20 .l:l. xc 1 i.e4 21 ll:\b5 .l:l. xc l 22 xc1 i.xf3 23
gf ll:\c6, and Black realised his advantage in the ending; Inkiov
Lputian, St John 1 988.
10
11
0-0
11t'aS
ll:\a6!?
A novel idea. The knight . heads for c5, so as to eliminate the

bishop on b3 in some variations. Instead, 1 1 . . . ll:\c6 1 2 h3 i.f5
produces a position which we have already examined in Game
No. 20. Gavrikov had introduced 1 1 . . . ll:\a6 a few months earlier,
but in a less prominent tournament than the USSR Championship.
Hence we have chosen Belyavsky-Gavrikov as the 'main' game.
12
ll:\eS
White utilises the fact that the e5-point is undefended. Stationing

his knight there, he takes aim at f7. An equal game results from
12 .i.c4 ( 1 2 a3 i.g4 1 3 h3 .l:l. ad8!) 12 . . . ll:\c5 1 3 a3 i.e6 1 4
4 i..f4 System 149

i.. x e6 ltJxe6 15 i.. e 5. In a later game Magerramov-Henkin,
Podolsk 1 989, White played 1l 1t' d4, and there followed : 12 . . .
J: d8 1 3 1t'e5 ltJd5 14 1t'g5 h6 1 5 1t'h4 ltJxf4 1 6 ef ltJc5 1 7 i.. c 2
i.. e6 (but not 1 7 . . . e6? 18 b4, and White wins) 1 8 f5 gf 19 b4
1t'c7? (a serious error; after 19 . . . 1t'a6! 20 be i.. x c3 21 i.. xf5
i.. g 7! 22 i.. x e6 1t'xe6 23 J: fe l , the game is about equal Magerramov. Now White obtains a strong attack) 20 be i.. x c3
21 J.xf5 i.. xf5 22 J: xc3 g7 23 J: e l J: d7 24 J: ce3 J: ad8 25
ltJe5 J: d 1 26 1t'g3 + i.. g 6 27 h4 1t'a5 (more stubborn resistance
was offered by 27 . . . J: xe 1 + 28 J: xe 1 h5 29 f4 1t'xc5 + 30 h2
e6 3 1 ltJxg6 fg 32 J: xe6) 28 J: xd 1 J: xd 1 + 29 h2 1t'xc5 30 J: f3 !
(30 h5? 1t' c l , and it is Black who wins) 3 0 . . . h 8 3 1 lll x g6+ fg
32 1t'b8+ 1 -0.
I2
ltJcS (97)
97
w
I3
i.. xf7 + !
From the material viewpoint this operation benefits Black, but

great complications now arise. A quieter move was 13 i.. c4.
I3
I4
IS
ltJxf7
b4!
J: xf7
xf7
The consistent follow-up to White's idea.

IS
I6
I7
lll dS
1t' xdS+
1t'xb4
ltJxdS
ltJe6
Better than 17 . . . i.. e 6 18 1t'xc5 1t' xc5 19 J: xc5 i.. xa2 20 i.. e 5!
b6 2 1 J: c2 i.. b 3 22 J: c3 i.. x e5 23 .l:t xb3 aS 24 f4 i.. g7 25 J: xb6
a4 26 J: f2 a3 27 J: a2 i.. b 2, with advantage to White (Gavrikov).
IS
J: c4
Wb2 (98)
150 4 j_f4 System

98
w
19
J:t fcl?
The black pieces are rather awkwardly placed, which compen

sates for White's material deficit. The position may be assessed as
double-edged with approximately equal chances, for example : 19
J:t c7! (threatening 20 'it'c4) 19 . . . f8 20 J:t xc8 + J:t xc8 21 'it'xe6
J:t c l 22 g4, when the resulting situation is completely calm. 19
j_gJ was also possible. The move played allows Black to extricate
himself at once.
j_d7!
19
The point is that after 20 'it'xd7 J:t d8 21 'it'a4 b5!, Black wins. If
White had played 19 ..tg3, he could have answered 19 . . . ..td7
with 20 'it'xd7 J:t d8 21 J:t f4 + . However, in that case Black has 1 9
. . . ..tf6, and i f 2 0 J:t fc l , only then 20 . . . ..td7!
20
h3
White loses not only after 20 'it'xd7, but also after 20 J:t 4c2
..tc6!
ll d8
20
20 . . . ..tc6 is even stronger; then 2 1 J:t xc6? be 22 J:t xc6 fails to

22 . . . 'it' b 1 + 23 h2 'it'f5.
21
'it'aS
Interestingly enough, this very position arose in the first game
that featured 1 1 . . . a6 : Lukacs-Gavrikov, Debrecen 1 988 (we
come to it at last). In that game White chose a different queen
move : 2 1 'it'f3, but after 2 1 . . . e8 22 J:t 4c2 'it'b5 23 ..tg3 ..tc6
24 'it'g4 'it'f5 25 'it' b4 J:t d3 26 J:t c4 'it'd5 27 e4 'it'b5, he had to
resign. Belyavsky presumably wasn't acquainted with that game;
at any rate, the move he plays leads to the same result.
a6
21
22
..tc7
J:t c8
4 i/4 System 151

23
24
25
26
.l:t 4c:Z
'irxb5
ib6
.l:t xc6
'ir b5
ixb5
.l:t c6
ixc6
Black's endgame advantage is obvious, and he exploits it with

no particular trouble - although it takes a full thirty moves.
27 f3
32 d:Z
c7 37
42 at
47 .tc3
ll:lf8 28 f:Z ll:ld7 29 ia7 ie5 30 e4 .td6 31 e3 e5

e6 33 .te3 .te7 34 c:Z b6 35 b:Z d6 36 .l:t d l +
h4 a5 38 h5 g5 39 h6 ll:lf8 40 .td:Z ll:le6 41 .tc3 .td6
ll:ld4 43 bl ll:le:Z 44 .tb:Z b5 45 .l:t d:Z ll:lf4 46 a3 ll:le6
b4
Slowly, but surely, Black accomplishes his task.

48 ab ab 49 : xd6 xd6 50 .txb4+ d7 51 c:Z ll:lf4 52 g4
ll:le:Z 53 .tc5 .tb5 54 d:Z ll:lf4 55 e3 .te:Z 56 .tb4 .tdl 0-1
Fianchetto System
Game No. 22
Karpov-Timman
Candidates Final (2nd game)

Kuala Lumpur 1990
The King's Fianchetto System (g2-g3, .ifl -g2) occurred twice
in the London/Leningrad match and twice in Seville. The exchange
in the centre (c4xd5 c6xd5) which took place in those games has
recently become highly popular; it follows that this 'symmetrical
variation' is worthy of close examination. In the present book I
have included my three most recent games with it. Two of them
were played in what may be considered the highest-calibre event
in the interval between the last two contests for the world crown :
my Candidates Final match with Jan Timman. In those two games
I played White, yet in the third example I had the black pieces.
With the large amount of supplementary material incorporated
in the notes, what is offered here is the most up-to-date survey of
the variation in question.
1
2
3
4
d4
c4
.!0(3
g3
.!0(6
g6
.ig7
c6
The advance . . . d7-d5, which actually defines the Griinfeld

Defence, is postponed by one move; in the event of an exchange
in the centre, Black is preparing to recapture on d5 with the pawn.
The other popular line, 4 . . . d5 Sed .!Oxd5, will be considered in
Game No. 25.
5
.i g2
d5
6
cd
On a previous occasion I decided to avoid exchanging in the

centre. Here is what happened : 6 .!Oc3 0-0 7 1Wb3 e6 8 0-0 .!Obd7
9 .if4 .!Ob6?! (9
b6 is more appropriate) 1 0 c5 .!Oc4 1 1 1Wc2
. . .
Fianchetto System 153

ltlh5 12 b3 ltlxf4 1 3 gf ltla3 14 1Wd2 b5 (on 14 . . . b6, White has
1 5 1Wb2 be 16 1W xa3 cd 1 7 ltla4 d3 1 8 .l:l. ac 1 de 19 .l:l. fe 1 , or 1 5
ltla4 b5 1 6 ltlc3, with the better chances) 1 5 .Z:. fe 1 .l:l. b8 1 6 h 1
a5 1 7 e3 f5? ( 1 7 . . . b4 1 8 ltla4 ltlb5, or 1 7 . . . 1Wc7 followed by . . .
f6 and . . . e5, was safer) 1 8 .tfl .td7 (now 1 8 . . . b4 is less good :
1 9 ltla4 ltlb5 20 ltlb6 ltlc7 21 a3) 1 9 .te2 .tf6 20 .l:l. g 1 h8 21
.l:l. g3 1We7 22 : ag1 .Z:. g8 23 1W c l ! b4 24 ltla4 .Z:. g7 25 1Wfl .l:l. bg8
26 1Wh3 .te8 27 ltlb6 1Wd8 28 1Wh6 .l:l. c7 29 ltle5 .l:l.cg7 30 .td3
ltlb5 31 .txb5 cb 32 f3 .th4 (he can't save himself with 32 . . .
.te7 33 ltlbd7! J.xd7 34 .l:l. xg6 J.f6 35 .Z:. xg7 .l:l. xg7 36 .Z:. xg7
J.xg7 37 ltlf7 + g8 38 1W xg7 + ri;xg7 39 ltlxd8) 33 .l:l. h3
J.f2 34 .Z:. xg6 J.xe3 35 .Z:. xg7 .Z:. xg7 36 .l:l. g3 1We7 37 .l:l. xg7 1W xg7
38 1Wxg7 + <J;xg7 39 c6 J.xc6 40 ltlxc6 .txf4 41 ltld7 1 -0;
Karpov-Kir. Georgiev, Wijk aan Zee 1 988.
Notwithstanding my success in that game, I still believe that
the immediate exchange on d5 promises White a more substantial
initiative.
6
7
8
ltlc3
ltle5
cd
o-o
e6 (99)
99
w
This book was virtually completed when one of the strongest

tournaments in chess history took place : Linares, 1 99 1 . The game
Karpov-Gelfand from that event is of interest for the theory of
the line we are examining. Play proceeded : 8 . . . J.f5 (a rare move
which 'almost' enabled Black to equalise) 9 0-0 ltle4 10 .te3
ltlxc3 1 1 be ltlc6 12 1Wb3 e6 1 3 ltlxc6 be 14 1Wa3 .l:l. e8 1 5 J.f4 e5
16 J.xe5 .txe5 1 7 de .l:l. xe5 1 8 e3 .te4 1 9 .Z:. fd 1 1Wf6 20 .l:l. d4 h5
2 1 h3 .:t b8 22 : ad 1 J.xg2 23 <J;xg2 1We7 24 flxe7 : xe7 25 .1:1. 1d2
154 Fianchetto System

l:t b6 26 c4 de 27 l:t xc4 (for twenty moves we have been very close
to a draw, but White has nonetheless extracted everything possible
from the position; exchanging most of the pieces, he has obtained
a superior rook ending. The remainder of the game provides a
good illustration of 'exploitation technique') 27 . . . g7 2S l:t dc2
l:t c7 29 g4 hg 30 hg f6 3 1 g3 e6 32 a4 d7 33 g5 : a6
34 l:t d4 + eS 35 l:t c5 l:t b6 36 f4 l:t d7 37 : xd7 xd7 3S
e5 e7 39 f4 : b4 40 : as : b7 4 1 e4 : c7 42 : c5 : cs 43
: c3 : eS 44 : c4 : cs 45 :b4 : c7 46 a5 d7 47 : b3 e7 4S
a6 d7 49 f6 cS 50 l:t h3 l:t d7 51 f5 gf 52 ef c5 53 : c3 : c7
54 g6 fg 5 5 fg d7 56 g7 : cs 57 l:t g3 1 -0.
9
.tgS
Postponing kingside castling for a while. For 9 0-0, see Game
No. 23.
9
1Wb6
9 . . . h6 weakens Black's position and wastes time. In Gutman
Zysk, Biel 1 9SS, White acquired a big advantage after 10 .te3!
1Wb6 1 1 1Wd2 h7 12 0-0 tt:Jc6 1 3 : fc l .td7 1 4 tt:a4 1Wc7 1 5
tOeS : adS 1 6 b4! 1W b6 1 7 b5! tt:Je7 l S a4.
10
1Wd2
tt:Jfd7
In Nikolic-Nunn, Amsterdam 1 9SS, Black equalised after 10
tt:Jbd7 11 .t e3 tOeS! 12 f4 t0d6. Gutman recommends 11 h3!
h6 ( 1 1 . . . tOeS 12 .te7, or 1 1 . . . tL!xeS 1 2 de t0d7 1 3 .te3! and
f4) 1 2 .te3, with the better chances for White.
The immediate 10
t0c6 has also been seen : 1 1 tt:lxc6 be ( 1 1
. . . 1W xc6 is also playable : 1 2 :t e l 1Wd7 1 3 0-0 b6 14 .th6 .tb7
15 : c2 : acS 16 : fc l t0e4, and Black maintains the balance;
Cvitan-Zysk, West Berlin 1 9SS) 12 0-0 t0d7 1 3 l:t fd l : bS 14 b3
f6 (after 14 . . . cS 15 de, or 14 . . . e5 15 de, White will play 1 6 .te3
and .td4, with obvious positional gains) 15 .th6 (here too 1 5
.te3 is not bad) 1 5 . . . .txh6 1 6 1Wxh6 cS 1 7 t0a4. This position
arose in Haritonov-Ivanchuk, Frunze 1 9SS. In Haritonov's
opinion, after 1 7 . . . 1W b4 or 1 7 . . . 1Wa5, White has a slight
advantage. In the game, Black played 1 7 . . . 1Wd6?, which could
have had severe consequences after l S e4!. Instead, White replied
l S de tt:lxc5 19 tt:lxc5 1Wxc5 20 : ac l 1Wd6 21 1W d2 f5 22 1Wd4 : f7
23 e3, and still preserved some positional advantages. In the end,
Ivanchuk was unable to hold the position, and lost.
11
.tel
I had already employed the manoeuvre .tcl -g5 against the
. . .

Dutch grandmaster four years earlier, but at this point I withdrew
my knight to f3.
Karpov-Timman, Bugojno 1 98 6 : 1 1 ltlf3 ltlc6 12 : d l ltlf6 ( 1 2
. . . 1t' b4 i s more precise) 1 3 0-0 .i.d7 (and here 1 3 . . . ltle4 was
better; then 14 ltlxe4 de 1 5 ltle5 ltlxe5 1 6 de .i.xe5 1 7 .i.xe4 1t'xb2
18 1t'xb2 .txb2 19 : b l gives White no more than a minimal
endgame advantage) 14 .i.xf6 .txf6 1 5 e4 1t'a5 1 6 1t'f4 (at this
point 16 ed ed 17 ltle5! ltlxe5 1 8 ltlxd5 1t'xd2 1 9 ltlxf6 + g7 20
: xd2 ltlc4 2 1 ltlxd7 ltlxd2 22 : d l : fd8 23 .txb7 would have
given White a large plus) 1 6 . . . .i.g7 1 7 : re t : ad8 1 8 ed ed 1 9
ltle5 .te6. The chances are now equal; subsequently w e both
made plenty of mistakes. At first I landed in a difficult position,
then Timman gave me the opportunity to restore equality and
acquire a decisive initiative.
11
ltlc6
The exchange on e5 promises nothing good : 1 1 . . . ltlxe5 1 2 de

1t'a5 1 3 f4 ltlc6 1 4 0-0 : d8 ( 1 4 . . . f6 1 5 ef .i.xf6 1 6 .tf2) 1 5 .tf2
.td7 16 a3! 1t'a6 1 7 b4 ltle7 1 8 : fd l , with a clear advantage to
White; Nikolic-Nunn, Brussels 1 988.
12
ltlxc6
be (100)
Better than 1 2 . . . 1t'xc6 1 3 .i.h6 .i.xh6 14 1t'xh6 1t'd6 1 5 h4!
and Black is in a dangerous position; Shpilker-A. Kuzmin,
Moscow 1 986.
100
w
13
:ct
I n Nikolic-Korchnoi, Amsterdam 1 988, White played the hasty

13 h4?!, when Black could have gained the better chances with 13
: b8 14 : b 1 c5 1 5 ltla4 1t'b5 16 ltlxc5 ltlxc5 1 7 de d4. But

then, the game continuation 13
a5 14 h5 .i.a6 1 5 : d 1 : fb8
. .
. . .

16 hg hg 1 7 b3 '1Vb4 1 8 fl c5 also gave Korchnoi a plus. There
followed 1 9 J:l. h4 c4 20 ..th6 ..th8 2 1 be 'lfxc4 22 ..tf3 J:l. c8 23
J:l. c l '1Vb4 24 g2 J:l. c4 25 a3 'lf xa3 26 J:l. ch l 'lfxc3 27 'lf xc3 J:l. xc3
28 ..td2 ..txd4 ! and Black won.
13
'1Vb4!
A good move. If at once 1 3 . . . J:l. b8, then 1 4 ll:la4 '1V b4 15 b3

'lfxd2 + 16 xd2, and White's chances are better, for example :
1 6 . . . J:l. b4? 1 7 J:l. xc6! ..txd4 1 8 ..txd4 ll xd4+ 1 9 c3 J:l. g4 20
J:l. c7 etc.
14
0-0
14 a3? would be premature : 14 . . . '1Vb3 1 5 'lfc2 J:l. b8 1 6 ll:la4

..ta6! with the initiative.
J:l. b8
14
15
b3
c5
Ridding himself of the backward pawn while he still can. After

1 5 . . . ..ta6 1 6 J:l. fd l ! J:l. fc8 1 7 ll:la4, Black has difficulty freeing
himself.
16
J:l. fd 1
Of course not 1 6 de?, because of 1 6 . . . d4!

16
17
..txd4
cd
..txd4
After 1 7 . . . e5? 1 8 ..txa7 J:l. b7 19 ..te3 d4 20 ll:ld5!, or at once

1 9 ll:lxd5!, Black would be left a pawn down.
18
19
20
'lf xd4
J:l. xd4
J:l. dd1
'lf xd4
ll:lb6
Given White's chosen plan, this is indispensable. 20 e4? fails to

20 . . . e5! 21 J:l. dd 1 d4, with good play for Black. If 20 e3, then after
20 . . . ..ta6 2 1 ll:la4 ll fc8 22 ll:lc5 ..tc4! Black similarly has an
easy game.
20
21
..tb7
e4!
Seeing that White's 'heavy artillery' is better mobilised, his

decision to open lines in the centre makes perfectly good sense.
21
22
23
de
ll:lxe4
J:l. c5
g7
Invasion of the seventh rank would be illusory : 23 J:l. c7 ll:ld5

24 ll d7 ll:lb6.
23
J:l. fd8

24
25
26
: xd8
lDc3
xg2
: xd8
J.. x g2
: c8?
White's positional advantage lies in his queenside pawn majority,
and Black has to play very accurately to avoid reaching a
dangerous ending. The correct move here was 26
ll d2!, and if
27 : c7 : cl 28 lDb5? : xc7 29 lDxc7 f6, Black is not at all
worse. A more precise line for White is 28 a4! a5 29 lDb5 : xc7
30 lDxc7 f6 31 lDa6 lDd5 32 lDc5 lDb4! 33 f4 (33 f3 e5!)
33 . . . lDc6 34 f3 e7 35 e4 d6 36 lDb7+ c7 37 lDc5
(37 lDxa5 f5 + !) 37 . . . d6, with a draw.
27
: xeS
lDxc8 (101)
. .
101
w
In this situation White could have utilised his active king

position and obtained substantial winning chances by playing 28
..tf3! In view of the threat to penetrate to the queenside, Black's
most natural course is: 28 . . . f5 29 ..te3 e5, and now
White has the very strong 30 lD b5! , preventing the stabilising
manoeuvre . . . lDc8-d6. It is by no means simple for Black to
defend. The variations I shall now quote were indicated by
Holmov. 30
..tf6 31 <.Pd3 <.P gS ! (Black's best chance is to
activate his king. He now threatens to advance by the route
g4-h3 or g4-f3, so White has to take prophylactic measures; but
this will lead to a weakening of his kingside, giving Black further
possibilities for counterplay. Instead, the pawn endgame after 3 1
. . . <.Pe6 3 2 <.Pc4 lDd6 + 3 3 lDxd6 <.Pxd6 3 4 b4! i s obviously lost
for Black, and White can also count on success after 32 . . . g5 33
a4) 32 h3! h5 33 <.Pc4 h4 34 ..td5 hg! 35 fg f4! (after 35 . . . e4 36
<.PeS! Black's defence would be incomparably more difficult) 36
gf+ <.Pxf4! 37 lDc3 (the straightforward 37 ..tc6 e4 38 <.Pb7?
. .

10d6 + ! would even lose). Now Black has to play very precisely :
(a) 37
10b6+? 38 >c6! (38 >e6? e4 39 >f6 e3! 40 >xg6?
>f3 4 1 h4 10d5! etc.) 38 . . . e4 (38 . . . g3 39 >b7 >xh3 40
>xa7 ll:ld7 4 1 b4 g5 42 a4 g4 43 a5 g3 44 b7 g2 45 ll:le2, and
White wins) 39 >b7 g3 40 >xa7 ll:ld7 41 b4! f3 42 a4 e2 43
ll:lxe2 >xe2 44 aS >d3 45 >b7 >c4 46 a6 bS 47 a7 ll:lb6 48
h4! and wins.
(b) 37 . . . gS! enables Black to hold the position : 38 >c6 >e3!
39 >b7 ltld6+ 40 >xa7 d2 4 1 ltld5 e4 ! (not 4 1 . . . ltlb5 + 42
>b6 ltlc3 43 >c5! e4 44 >d4! ltlxd5 45 >xe4 ltlc3 + 46 >f5
and White wins but 43 . . . ltlxa2! draws-G. Flear) 42 b6! (less precise
is 42 a4? >c2 43 b4 b3 44 a5 c4 45 ll:le3 + >xb4 46 b6
ltlc4+ with a draw) 42 . . . e3 43 ltlxe3 >xe3 44 >c6 .!ticS 45
>d7 ltlb6?! (45 . . . ltla7? is inferior: 46 a4 d4 47 >e6 c3 48
>f5! xb3 49 xgS >xa4 50 h4 ltlc6 5 1 >f6! and wins; or
47 . . . >e4 48 b4! ltlc6 49 bS ltla5 50 b6 f4 5 1 >d5 >g3 52
>c5 >xh3 53 bS, again winning for White) 46 >e6 e4! 47
a4 ltld5 48 a5 ltlb4 49 >d6 >d4!, or 49 >f6 >f4, with a draw
in either case.
These variations show that although Black achieves a satisfac
tory result, it is only with immense difficulty.

28
f4?
This throws away White's advantage, since the exchange of a

pair of kingside pawns increases Black's chances of a peaceful
outcome.
28
29
fS!
ltla4
Having let my winning chances slip, I commit a further

inaccuracy. 29 f3 was simpler.
29
30
>f6
.!ticS
Preventing 30 . . . e5 on account of 3 1 ltld7+.

30
31
32
ltld3
ltlb6
ltld7
ltlb4
I should have brought the king forward : 32 >f3 e5 33 fe +
ltlxe5 + 34 ltlxe5 >xe5 35 h4! with a draw.
32
33
ltlc6
34
35
ltld4
fe +
eS
a6
ltl xe5

The pawn ending would be lost for White.
35
36
e7
n
36 f2 is answered by 36 . . . etlg4 + .
36
37
38
39
40
d6
e2
etlcl
dS
e4
a4
b4
.!Of3
etld4+
In the event of 40 . . . .!Oxh2!, White could maintain the balance

with 41 b5 ab 42 ab d5 43 .!Od4! .!Og4 ! 44 b6 d6 45 .!Oe6!
c6 46 .!Of8 .!Of6 47 f3 xb6 48 f4 c6 49 g5 .!Oe4+
(or 49 . . . tt!h5 50 h4 d5 51 .!Oxh7) 50 f4 d6 51 lt!xh7
e7 52 g4; alternatively 42 . . . lt!g4 43 b6 tOeS 44 b7 lt!d7 45
ltlb4 lObS 46 .!Od3! and again White draws, as Black cannot win
without the use of his knight.
41
42
lt!xd4
bS
xd4
t- t
After 42 . . . ab 43 ab c5 44 e3 xb5 45 f4 c4 46
g5 d3 47 h6 e3 48 h4! f3 49 xh7 xg3 50 xg6,
both sides' resources are completely exhausted.
Game No. 23
Karpov-Timman
Candidates Final (4th game)

Kuala Lumpur 1990
1
d4
lt!f6
2
3
4
s
6
7
8
9
c4
lt!f3
g3
J.g2
cd
.!Oc3
tOeS
g6
J.g7
c6
dS
cd
o-o
e6
o-o
In our discussion of this game, we shall scrutinise the fine points

of the opening play arising from White's 9th move. The middlegame
will proceed at a slow pace, yet the ending (after the adjournment)
will be exceptionally interesting, and although this does not strictly

belong to the subject of the book, we shall give a large number
of entertaining endgame variations.
lt)fd7
9
The best move. After 9 . . . lt)c6 10 li)xc6 be 1 1 a4 lt)d7 1 2
.tf4 White has a clear positional advantage; Akhmilovskaya
Chiburdanidze, Women's World Championship match 1 986.
10 f4 (102)
102
B
Karpov-Kasparov, 3rd game, London 1 986, continued differ

ently : 1 0 lt)f3 (the exchange 10 lt)xd7 .txd7 gives White nothing :
1 1 e3 lt)c6 1 2 b3 "fle7 1 3 .tb2 .C. fc8 t-t ; Portisch-Nunn, Budapest
1 987) 10 . . . c6 1 1 .tf4 lt)f6 1 2 lt)e5 .td7 (after 1 2 . . . li)xe5 1 3
.txe5, White's chances are to be preferred) 1 3 "fld2 li)xe5 1 4
.txe5 .tc6 1 5 .l:lfd 1 ( 1 5 .C. ac 1 , with the prospect of f2-f3 and
e2--e4, would have preserved the initiative) 1 5 . . . lt)d7 1 6 .txg7
q;xg7 1 7 .C. ac 1 f6 1 8 "flf4 "fi bS 19 "flxb8 .C. axb8 20 f3 .C. fd8 2 1
q;f2 .C. bc8 22 e 3 lt)e8 23 .C. d2 lt)d6 24 .l:l dc2 f8 2 5 .t n e7
26 .td3 f5 27 h4 h6 28 b3 (in his notes to the game, Kasparov
states that after 28 g4 Black would be in quite a dangerous
position. But it seems to me that the continuation 28 . . . .C. f8 29
g5 e4 + 30 .txe4 fe 3 1 f4 h5, or 30 fe fe + 31 q;e2 ed + 32
q;xd3 hg 33 hg .C. f5!, promises White nothing) 28 . . . g5 2 lt)e2
.td7 30 .C. c5 b6 3 1 .C. c7 .C. xc7 32 .C. xc7 .C. a8 33 lt)g1 e8 34 .C. c 1
.C. c8 35 .C. xc8 t-t .
10
lt) c6
In game 1 3 of the 1 986 match, my opponent drove tbe knight
back with 10
f6, but after 11 lt) fJ c6 12 .te3 b6 l 3 .tf2
f5 14 li)e5 .td7 1 5 "fld2 li)c8 16 "fle3 q;hs 1 7 .C. fd 1 ( 1 7 .C. fc l was
. . .

stronger) White generated considerable pressure - although the
game was eventually drawn after blunders by both sides.
White can retain the initiative by retreating his knight to d3.
The game Hulak-H. Olafsson, Wijk aan Zee 1987, is instructive :
( 1 0 . . f6) 1 1 o!Lld3 o!Llc6 1 2 e3! (more precise than 1 2 .i.e3 o!Llb6 1 3
b3 .i.d7 14 o!Llc5 J:l b8 1 5 "it'd2 f5 1 6 J:l fc 1 .!Llc8 1 7 .i.f2 o!Lld6, with
equality; Nikolic-Nunn, Linares 1 988) 1 2 . . . f5 (Andersson-Nunn,
Brussels 1 988, is worth mentioning : 12 . . . o!Llb6 1 3 b3 ..td7 1 4
..t a 3 J:l e8 1 5 "it' d 2 o!Ll e 7 16 o!Ll c 5 J:l b 8 1 7 J:l fe 1 f5 1 8 .i.fl o!Llec8
19 J:l ac1 .i.c6 20 o!Lld3 o!Lld6 21 o!Lle5 ..tf8 22 "it'b2! J:l e7 23 J:l c2
J:l c7 24 J:l ec1 J:l bc8, and now instead of 25 o!Llb1, which led to
approximate equality, White could have played 25 .i.c5!, when
according to Nunn the threat of 26 "it' a3 gives him a noticeable
plus) 1 3 .i.d2 (a good alternative is 1 3 o!Lle5 o!Lle7 14 b3 o!Llf6 1 5
..ta3 ..td7 1 6 J:l c 1 J:l e8 1 7 J:l f2 o!Lle4 1 8 J:l fc2 o!Llc6 1 9 .i.fl a 6 20
o!Llxe4 fe 21 ..td6 .:l c8 22 h4 o!Llxe5 23 de "it'aS 24 hS; Dzhind
zhikhashvili-Mestel, Reykjavik 1 990) 1 3 . . . o!Llf6 14 J:t c l ..td7 1 5
o!Lle5 J:l e8 1 6 h 3 o!Llxe5 1 7 de o!Lle4 1 8 o!Llxe4 de 1 9 "it' b3 .i.c6 20
..tb4, with advantage to White.
Shortly after the London/Leningrad match, Kasparov played
this variation himself with White, against Nunn (Brussels 1 986).
The exchange on e5 unexpectedly led to a crushing defeat for
Black : 10 . . . o!Llxe5 1 1 fe o!Llc6 1 2 e4! de 1 3 ..te3 (103)
.
103
8
With Black at the moment a pawn up and not having made

any noticeable mistake, it is hard to imagine that he will resign
after six ( ! ) more moves. 1 3 . . . f5 14 ef J:l xf6 (the play up to here
was still on familiar lines. For example, this position had arisen
by transposition in Sveshnikov-Mikhalchishin, Lvov 1983; after

1 S .J:I. xf6 .txf6 16 ll:le2 ll:lb4, White's advantage evaporated.
Kasparov doesn't waste time exchanging rooks) 1 S ll:lxe4 .J:I. xfl +
1 6 1Wxf1 ll:lxd4? 1 7 .J:I. d 1 eS 1 8 lOgS, and Black stopped the clock.
The following beautiful variation is possible : 1 8 . . . 1We7 1 9 .tdS+
.te6 20 l:l xd4 ed 2 1 .txe6+ *h8 22 lll f 7+ *g8 23 /Od8+
*h8 24 .tgS! 1Wb4 2S /Of7+ *g8 26 ll:leS + *h8 27 /Oxg6+
hg 28 1Wh3 + . Incidentally, all this actually happened in a corre
spondence game Hjorth-M. Andersson ( 1 986), and at this point
Black resigned.
Instead of 1 6 . . . ll:lxd4, Black might play 1 6 . . . .txd4 1 7
.txd4 /Oxd4 1 8 l:l e l eS 1 9 1Wf6. Kasparov judged this position to
be in White's favour, in view of 19 . . . 1W xf6 20 lll xf6 + *g7 2 1
/Oe8 + *f8 2 2 .J:I. xeS with the better ending, but Gutman has
suggested 19 . . . 1!fb6 20 *hl .td7 21 1!fxe5 l:l f8 22 lOgS .tc6.
However, after 23 .tdS + ! (G. Flear) Black can resign!
12 e4, then, seems quite strong but White can also play the
solid 1 2 .te3!?. This position arose in Karpov-Timman,
Amsterdam 1 986, though by a different move-order ( 1 0 . . . /Oc6
1 1 .te3 ll:lxeS 12 fe). Let us look at a few moves of that important
encounter.
12 .te3 f6 13 ef .J:I. x/6
Better than 1 3 . . . .txf6 1 4 1Wd2 .td7 1 S * h 1 .J:I. f7 1 6 .tg1
.te8 1 7 l:l ad 1 .tg7 1 8 l:l xf7 .txf7 1 9 e4, with the advantage;
Makarov-Glek, Moscow 1 986.
14 1Wd2 .td7 15 *h1 l:l xf1 + 16 l:l xfl 1We7 (104)
104
w
This is a safer post for the queen than aS ( 1 6 . . . 1WaS 1 7 a3 l:l f8

1 8 l:l xf8+ .txf8 1 9 .tg 1 ) or b6 ( 1 6 . . . 1Wb6 1 7 .tg1 /Oxd4? 1 8
l:l f4).

1 7 '4 d1
Over-protecting the d-pawn and preparing ..te3-g1 and
e2-e4. But White can also withdraw his bishop to g1 at once.
Ribli-Nunn, Dortmund 1 987, went 1 7 ..tg1 '4 d8 18 a3 h8? ( 1 8
. . . ..tc8 1 9 ..te3 '4 f8 i s safer) 1 9 e4! (the fiasco against Kasparov
has not put Nunn off this variation, but once again the advance
of the centre pawn puts him in a difficult position) 19 . . . de 20
llJxe4 ..tc8 2 1 ..te3 (21 llJg5 ..tf6) 2 1 . . . .l:l8 22 : xf8 + 1Fxf8 23
b4, with a clear advantage.
17 . . . '4 c8
In Karpov-Chiburdanidze, Bilbao 1 987, the Women's World
Champion played 17 . . . h8 18 a3, and only then 1 8 . . . : c8.
After 1 9 .*.g5 1Ff8 20 :n 1Fg8 21 e3 h6 22 ..tf6 .*.xf6 23 : xf6
: f8 24 : f2 1Fg7 25 : xf8 + 1Fxf8 26 e4 de 27 lt:lxe4 b6 28 g1
'ik g7 29 d5 ed 30 1Fxd5 1Fd4+ 31 1Fxd4 + llJxd4, the game had
turned from an opening into an ending, with White retaining a
minimal edge. But Maya defended precisely, and obtained a draw
in spite of all my efforts.
18 a3!
White's intention is to retreat his bishop to g1 and seize the
centre. But in reply to the immediate 1 8 ..tg1 , Timman had
prepared 18 . . . 1Fb4 ! . Then after 19 e4 de 20 .*.xe4 ..te8, Black
can secure his position with . . . .*.e8-f7 (2 1 d5? ..txc3!). If instead
20 llJxe4 1Fxd2 2 1 : xd2, Black has the tactical device 2 1 . . . llJxd4
22 : xd4 (22 ..txd4 .*.xd4 23 : xd4 : c 1 + 24 ..tfl ..t b5) 22 . . .
..txd4 23 ..txd4 : c 1 + 24 ..tgl .*.c6, with counterplay in a
complex endgame.
18 . . . ..t/6
White easily refutes 1 8 . . . llJa5 with 1 9 llJxd5 ed 20 .*.g5 llJc4
21 ..txe7 llJxd2 22 .*.xd 5 + h8 25 : xd2.
19 .*.g1 ..tg5?
A more precise move was 19 . . . 1Fg7, hindering e2-e4 for the
moment, on account of 20 e4 de 21 llJxe4 ..txd4 22 .*.xd4 llJxd4
23 1Fxd4 1Fxd4 24 : xd4 : c 1 + 25 .*.fl ..tb5 ! .
2 0 1Ve1 llJd8 2 1 e4 de 22 1Vxe4 b 6 23 d5
Black could not prevent this breakthrough, and has to go over
to wholly passive defence. The results of the opening can now be
summed up as miserable for Black. The game, however, lasted
another 35 moves and ended in a draw.
11
.*.e3

At this point the exchange on e5 leads to variations we have
looked at already ( 1 1 . . . dxe5 12 fe f6 etc.). In the main game
we are examining, Black drives back the centralised knight and
follows with . . . b6. I should mention that on two occasions in
Seville, Kasparov played 1 1 . . b6 at once, and equalised easily
in both cases. We shall thoroughly discuss the immediate knight
move in the notes to Game No. 24.
.
11
f6
b6
12 d3
'fle7
13
b3
13
..td7 would transpose into the Nikolic-Nunn game that
we have already seen. So it may be said that we are only now
breaking new ground.
. . .
14
15
16
17
18
19
a4
.tel
e3
..ta3
l:t c l
..txf8
..td7
l:t fd8
..te8
'flf7
..tf8
'fi xf8 (105)
105
w
As the result of some complex manoeuvres, a certain advantage

for White has emerged on the queenside.
20 g4
'fie7
l:t ac8
21
'fld2
22 e2
l:t c7
23 l:t c5
c8
24
f5
gS
White is slightly better after 24 . . . d6 25 df4 (25 fe? e4 26

..txe4 de 27 df4 'fi xc5, and Black wins) 25 . . . gf 26 gf xf5 27
xd5 ed 28 : xf5.

25
g3
2S fe 1i'xe6 26 c3 (26 g3 g6 27 fS 8e7) 26 . . . 8e7 27

e4 is less clear.
Now the threat of bringing the knight to f5 is more dangerous :
2S . . . .i.f7 26 fe 11'xe6 (26 . . . xe6 27 h5 .l:l. f8 28 b4!) 27
f5. To avoid this, Black sacrifices a pawn.
e5!? (106)
25
106
w
26
Wet
An inaccuracy, committed after much thought. The sheer variety
of possibilities put me in rather a quandary, and I decided to
decline the pawn sacrifice although Black's position would almost
have become critical if I had calmly accepted it. After 26 xd5+
(26 .l:l. xd5 ed 27 ed .i.f7! 28 .l:l. xd8 + Wxd8 gives Black counterplay
for the pawn) 26 . . . h8, White has a wide choice :
(a) 27 xc6 J.xc6 28 1W c1 ed 29 e4 !, or
(b) 27 .l:l. fcl ed 28 e4! . This strategic operation involving the return
of the pawn and the inevitable break with e4-e5 secures White a
considerable plus.
(c) 27 de xeS 28 .l:l. xc7 1Wxc7 29 .l:l. c 1 'fle7 30 xeS 1W xeS 3 1 1W d4!
is also good.
(d) 27 1Wcl .l:l. xd5 28 .l:l. xdS b6 (28 . . . b4 29 l:t cS!) 29 de b4 30
xb4 .l:l. xcl 3 1 .l:l. xc l , or 27 . . . b6 28 .l:l. xc6 J.xc6 29 J.xc6 ed
(29 . . . .l:l. d6 30 dS .l:l. xdS 3 1 .i.xdS) 30 ed .l:l. xd4 (30 . . . .l:l. d6 31 dS
.l:l. xdS 32 .l:l. e 1 !) 3 1 .l:l. e 1 1Wd6 (3 1 . . . .l:l. xd3 32 .l:l. xe7 xe7 33 Wet
xc6 34 1We8 + g7 3 S hS+ h6 36 1Wf8 + and mates) 32
.l:l. e8 + g7 33 l:t e6 - with a very strong initiative for White in
either case.
(e) 27 .i.g2 (in some of the foregoing variations the white king's

bishop departs from the board, but it is also possible to bring it
home before anything else) 27 . . . ed 28 e4!, followed by e4-e5 (if
necessary supported by a rook - : fe l ). If 28 . . . lt:le5, then 29
: xeS! fe 30 f6 and 3 1 'ill xg5 is crushing.
The queen move to c1 allows the centre to be blocked; if l had
been playing for that, I should have continued 26 : fe l l e4 27
lt:lb4! - possession of the c-file is not without significance. The
trouble is that I had 'too many irons in the fire' (trying to control
the c-file and f-file at once). Now Black nullifies the danger in the
centre and obtains a fully viable game. The ensuing phase, as I
have said already, is a trifle dull; the most interesting events occur
only after the adjournment.
26
b6
27
28
29
30
: c2
lilf2
1i'd2
l:tfcl
e4
lt:ld6
.l: dc8
aS
Black also stands quite well after 30 . . . lt:la5 3 1 : xc7 : xc7 32

: c3 'ill d 8 33 'ill c2 : xc3 34 'ill xc3 'ill c8!
31
-*.fl
If White hadn't made the superfluous queen moves, this bishop
manoeuvre would now be extremely dangerous for Black.
The attempt to build up with 3 1 'ill c 3 would be countered by
3 1 . . . -*.d7 and 32 . . . lt:le8.
31
lt:lb4
3 1 . . . lt:lxd4! looked inviting, but after 32 ed e3 33 -.d3 ef+ 34

xf2 : xc2 + 35 : xc2 lt:le4+ 36 f3, the draw is not far away
(the exchange of the good knight on c6 for the bad one on f2 is
not advantageous to Black). The match situation compelled
Timman to maintain the tension in the hope of seizing the
initiative.
32
33
34
35
36
37
38
39
: c3
lild1
: xc6
: xc6
lilc3
f2
40
el
e1
d1
-.d7
: c6
: xc6
-.xc6
f8
e7
f8
-.cs
g7

Under pressure from the clock, we have both been marking
time with our kings a little, and I should mention that Black
missed quite a good opportunity : 40 . . . hS! 41 gh (41 h3 with
equality) 4 1 . . . lll xfS 42 .J.. h 3 .J.. d 7!
41
42
lll a 2
11rxa2
lll x a2
11rc7
A possibility was 42 . . . c3 + 43 f2 bS 44 ab lll x bS (but not

44 . . . .i.xbS 45 lll h S + f7 46 .i.xbS lll x bS 47 b4!).
43
44
45
46
f2
1i' b2
.tel
el
f8
e7
d8
c8
d2
ct
b7
11re7
46 . . bS!? is sharper.
.
47
48
For drawing purposes, 48 . . . 11rxc1 + was simpler.

.i.d7
49
el
50
51
Ill eS
f2
11Vhl ! (107)
107
B
I didn't feel like settling for a draw, and decided to provoke

mind-bending complications by abandoning my queenside to its
fate while going ahead on the kingside. Obviously there could be
no question of calculating the variations to the end.
51
52
53
54
55
h4!
hg
11rxh7
11re7
11V b4
11V xb3
fg
xa4
11rc6

56
57
58
59
1hg5
'fll e7
'fll xd6
.t. dl
a4
'fll d6
ltlxd6
.tb5?!
It was hardly a good idea to block the path of his pawn. After
59 . . . b5, it is most likely that the game would soon have been
drawn; at any rate Black would not be risking anything. Here are
some sample variations : 59 . . . b5 60 f6 (60 ltle2? b4 61 ltlc1 c7)
60 . . . .txg4 61 .txg4 ! a3 62 .te6 c6 63 ltle2 (not 63 f7 ltlxf7
64 ltlxe4 ltlg5! 65 it:lxg5 a2, and Black wins) 63 . . . a2 (63 . . . b4 64 ltlcl)
64 .t.xd5 + xd5 65 ltlc3 + c4 66 ltlxa2 b4 (nor does he gain
anything from 66 . . . b3 67 ltlc1 + b2 68 ltle2 b4 69 d5 b3
70 g3 c2 71 ltld4+ c3 72 ltlxb3 xb3 73 f4 c4 74
e5) 67 ltlxb4 xb4 68 g3 c4 69 f4 d3 (69 . . . d5
70 g5 e6 7 1 g6) 70 e5 ltlf7 + 71 e6 ltlg5 +, with a
draw.
60
61
ltle2
a3
ltlcl (108)
108
8
In this position Timman sealed his move after nearly half an

hour's thought. Many commentators were misled by Black's
queenside activity, and in the local newspapers the following day
there were even headlines like "Can Karpov survive?" In actual
fact, adjournment analysis revealed that the only winning chances
lie with White, whose pawns are much more dangerous than his
opponent's. All the same, we shall see that the rightful outcome
of this tense struggle would have been a draw.
61
c7
A good sealed move, but of course we had analysed others too.

Let us look at the main variations discovered by Zaitsev and
myself:
(a) 61
-*.d7 62 -*.b3! c6 63 g3 c4 64 f4 b2
(threatening the deadly . . . d3 +) 65 g5! c4 66 -*.xc4 (given
White's active king, the creation of another black passed pawn is not
dangerous) 66 . . . de 67 f6 b5 6S a2 b4 69 xb4 b5 70 a2
a4 71 c1 a2 72 xa2 b3 73 c1 + c2 74 e7 xc 1
75 f6 c 3 7 6 f 7 -*.xg4 77 fS(11t') c2 7 S 11t'f4 -*.f3 7 9 d5 d 2 S O d6
cl(11t') S 1 d7, and Black is defenceless.
(b) 61
..te8 62 a2 (after the careless 62 g3, the black
knight performs the same nimble triangulation that we have seen
before : 62 . . . c4 ! 63 f4 b2, and wherever the bishop goes,
Black plays 64 . . . d3 +) 62 . . . c6 63 -*.b3 c4 64 g5 d6 65
f6 -*.f7 66 g3 c4 67 f4 d6 6S c3 ..tg6 69 -*.xc4 de 70
b5 + c6 7 1 xa3 d5 (7 1 . . . b5? 72 e5 c3 73 c2 -*.f7
74 d5 + ! -*.xd5 75 g6) 72 b5 c6 73 c3 b5 74 e5 b4 75
d 5 + d7 76 a2 b3 77 c3 b2 7S d4, with a decisive
advantage.
(c) 61
c6 62 g3 c4 (if 62 . . . -*.c4 63 f4 d7 64 g5
a2 65 xa2 -*.xa2 66 g6, Black is mated after either 66 . . . eS
67 -*.a4+ e7 6S -*.xeS xeS 69 g5 b5 70 h6 b4 7 1 g7
f7 72 h7 b3 73 gS(11t')+, or 66 . . . e7 67 g7 f7 6S f6 -*.c4
69 -*.h5+ gS 70 -*.g4 -*.b5 71 -*.e6+ f7 72 f5 -*.eS 73
g6 b5 74 -*.xd5) 63 -*.e2 d6 (if 63 . . . xe3?, then 64 -*.xb5 +
xb5 65 f6, and the pawn goes on to queen. Other variations
are also unacceptable : 63 . . . d6 64 -*.xb5 + xb5 65 f4
b4 66 g5 c3 67 g6 e8 68 e5 b2 69 g7 xg7 70 f6
xc l 7 1 fg a2 72 gS(11t') a 1 (11t') 73 11t' g 1 + b2 74 11t'xa1 + xa 1
7 5 xd5, o r 6 3 . . . d6 64 -*.xb5 + xb5 6 5 a2! d6 66 f4
c7 67 g5 e8 68 g6 f6 69 g5 e7 70 b4 b5 7 1 c6 + !
d6 7 2 xf6 a 2 73 g 7 a 1 (11t') 74 gS(11t') xc6 75 11t'e6+) 64 f4
-*.a4 65 g5 b2 66 g6 d3 + 67 xd3! ed 68 ..txd3 e7 69 g7
f7 70 f6 g8 (70 . . . a2 7 1 -*.h7 a 1 (11t') 72 g8(11t')+ xf6 73
11t'f8 + e6 73 -*.f5 mate) 71 g5 a2 72 h6 a 1(11t'), and a
familiar mate follows : 73 -*.h7 + f7 74 g8(11t')+ xf6 75 11t'f8 +
etc.
Now let us see what happened in the game.
c4
62
g3
62 . . . d7? loses in a way which we have seen before : 63 f4
-*.c4 64 g5 a2 65 xa2 ..txa2 66 g6 e7 (66 . . . e8 67 -*.a4+)

. .
1 70 Fianchetto System
67 g7 f7 68 f6 .tc4 69 .tg4 g8 70 .te6 + t;)f7 7 1 f5 .tb5
72 g6 .te8 73 .txd5 b5 74 .ta2 b4 75 .tb3.
63 .te2
But not 63 f4? t;)b2! threatening . . . t;)d3 + .
.te8!
63
The strongest defence. 63 . . . li)xe3 is inadequate : 64 .txb5 t;)c2
65 g5 t;)xd4 66 f6 d6 67 g6 e6 68 g7 f7 69 .te8 + g8
70 f4, and wins.
64
f4
Podgayets revealed a striking variation culminating in a
problem-like mate : 64 g5 t;)xe3 65 f4 t;)c2 66 e5 t;)b4 67
f6 d6 68 g6 t;)c6 69 ..tb5 e3 70 g7 t;)e7 7 1 ..txe8! e2 72
li)xe2 a2 73 t;)c3! t;)g8 + 74 f7 t;)h6+ 75 f8 a l (V) 76 t;)b5
mate (109) .
109
B
It would indeed be pleasant to finish the game like this, but

unfortunately Black has a defence : 64 . . . d6 65 ..txc4 de 66
t;)a2 ..ta4 67 t;)c3 .tb3 68 t;)b5 + e7 69 li)xa3 c3 70 f4 b5!
71 xe4 b4 72 d3 ba 73 xc3 .td5 with a draw.
64
t;)b2
In my adjournment analysis I considered 64 . . . b5 to be more
precise : 65 t;)a2 t;)b2 (but not 65 . . . d6 66 g5 t;)b2 67 g6 b4 68
g5! b3 69 t;)cl .ta4 70 g7 a2 71 g8() a l (V), as White obtains
an irresistible attack with 72 Vb8 + d7 73 f6! Vxc l 74 ..tg4+
c6 75 V c8 + ) 66 f6 t;)d3 + 67 f5 b4 68 .td l . But the move
played is not yet the cause of a catastrophe, either.
65 g5
The sole winning attempt. It is true that after 65 t;)a2, the
following long variation gives White good prospects : 65 . . . d6
Fianchetto System 1 71
66 gS e7 (66 . . . lilc4 is a mistake in view of 67 ..txc4 de 68
f6 ..ta4 69 g7 ..tb3 70 f6 c3 7 1 lilxc3 a2 72 lilxa2 ..txa2 73
gS) 67 h6 f6 68 lilb4 ..tf7 (inadequate alternatives are 68 . . .
ll:ld3 69 lilxdS + f7 70 gS a2 7 1 g6+ f8 72 g7 + f7 73 ..thS+
g8 74 lll e 7 mate, and 68 . . . ..tc6 69 ..tfl ! lildl 70 gS + f7 70 . . . xf5 7 1 ..th3 mate, is pretty - 7 1 g6+ etc.) 69 ..tfl ! lild l
(it is too late for counterplay with 69 . . . lild3 70 gS + e7 7 1
lilc2 a 2 7 2 g7) 7 0 g5 + e7 7 1 g7 lilxe3 7 2 f6 + e6 73
..th3 + lilfS + 74 ..txfS + xf5 75 xf7 e3 76 g7! e2 77 ll:lc2
a2 78 f7 e 1 (1i') 79 lilxe l a 1 (1i') 80 f8(1i')+. However, after the
correct 65 . . . b5!, nothing can be found for White.
lild3
65
65 . . . lilc4 comes too late : 66 f6 lilxe3 67 e7.

al
66
lil b3! ?
67
68
ll:lal
69
-*.b3
bS
..tdl
b4 (IJO)
Bringing the king a little closer to the pawns does not work :
68 . . . d6? 69 ..tb3 lilcl 70 h6 b4 7 1 f6 e6 72 g7 lilxb3
73 lilxb3 ..ta4 74 lilcS + ! d6 75 f7 a 1 (1i') 76 f8(1i')+ c6 77
1i'c8 + d6 78 'ifd8+ c6 79 1i'd7 + b6 80 1i'b7 + aS 8 1
1i' a6 mate.
1 10
w
lilcl
After 69 . . . d6 70 f6, the g-pawn cannot be stopped.

70
..t xd S
d6
71
..tc4
..tbS
Not 41 . . . e7 42 f6 + f8 43 h6 ..tf7 44 dS b3 75 d6!, or

7 1 . . . ..ta4 72 f6 b3 73 h6! b2 74 f7 b1 (1i') 75 f8 ('if)+ d7
76 1Wf5 + c7 77 1i'a5 + , but here 74 . . . ba (11') leads to a
completely unclear position - G. Flear.
An interesting possibility is 71 . . . b3 72 ..txb3 lilxb3 73 lilxb3
d5 74 a1 c4 75 .tf6 c3 76 e7 b2 77 xeS xa1
78 f6 b2 79 f7 a 1 (W) 80 f8(W) Wa4+ 81 f7 b3 + . We had
reached this position in our analysis, and Zaitsev suggested 82
.tf6 W xe3 83 e5 .tc2 84 Wf4 .td3 85 d5 with chances of
success. But in the post-mortem, Timman pointed out the more
effective 82 d5! Wxe3 (82 . . . W xd5+ 83 'Oftg6 with a technically
won ending) 83 W b4 + ..tt c 2 84 g5 Wf3 + 85 g7 e3 86 Wc4+
b2 87 W b5 + c1 88 d6 e2 89 Wc5+ b2 90 Wb6+ c1 91
d7 e l (W ) 92 d8(W) Wec3 + 93 Wdf6 etc. According to computer
analyses, this ending is a win for White.
e7
72
..tg8
73
h6
After 73 f6 + f8 74 ..te6 b3 (74 . . . ..ta4 leads to the same

result as in the game) 75 .i.xb3 xb3 76 xb3 ..tc4 77 at f7
78 f5, White wins. However, Black has the saving move 74 . . .
.i.d7!, and if 75 ..txd7 (it is more sensible to settle for a draw)
75 . . . b3 76 f5 b2 77 g5 e2! 78 g6 g3 + ! 79 g5 h5! 80
xh5 ba(W) 8 1 g7 + f7 82 .i.e8 + (82 ..te6 + xf6 83 g8(W)
'if h 1 +) 82 . . . ..tt xf6, it is Black who unexpectedly wins.
f8?
73
It is only now that Timman commits the decisive error. He
could have drawn with 73 . . . b3! 74 .i.xb3 xb3 75 xb3 .i.c4
76 a l f6!, as pointed out by his second, the Hungarian
grandmaster Sax.
.i. d7
74
..te6!
75
76
gS
g6
b3
1-0
Black resigned rather than be mated : 76 . . . ..txe6 77 fe b2 78

g7 + rJ;e7 (78 . . . g8 79 e7 f7 80 g8(W)+ xe7 8 1 Wg5 +
rJ;f7 82 Wd5+ rJ;f6 83 'ifc6+) 7 9 g8(W) ba(W) 8 0 W f 7 + d6 8 1
W d 7 mate.
After this victory, it became clear that another duel for the
world crown between Kasparov and myself was inevitable.
Game No. 24
Wojtkiewicz-Karpov
Haninge 1990
I have played the White side of the Griinfeld so often, especially
the g2-g3 system, that when the symmetrical variation arose in
the present game, it was with some interest that I handled it for
the other colour.
lt:l f6
1
lt:lf3
2
3
4
5
c4
g3
.i.gl
d4
g6
.i.g7
0-0
c6
I have rarely played either the King's Indian or the Griinfeld

with Black. But given the choice between them, I prefer the latter.
6
7
8
9
10
11
lt:lc3
cd
lt:l e5
0-0
f4
.i. e3
d5
cd
e6
lt:lfd7
lt:lc6
lt:lb6 (1 1 1)
111
w
Kasparov had played this knight move against me in the first

and third match game in Seville, and now I decided to use it
myself. An alternative of equal value is 1 1 . . . f6 (see Game No.
23). In the present game, you might say that I combined the two
ideas.
Before proceeding with the main game, in which White played
1 2 b3, let us look at the two encounters from Seville, where I
preferred 1 2 .i.f2. The first match game went as follows :
12 j_f2 .i.d7 13 e4 l0e7 14 l0xd7
In the game Drasko-Nikolic, Vrnjacka Banja 1 987, where it
appears this position occurred for the first time, White opted for
14 ed lt:lbxd5 1 5 lt:lxd5 lt:lxd5 1 6 Wb3 .i.c6 1 7 : ac l 1Wa5 1 8 .:l. c5
'it'a6, and a draw was agreed.
Another possibilty is 14 a4! de 15 aS .!Dbd5 16 .!Dxe4 .:t b8 17
'iVb3 ..te8 18 .l:tfcl .!Dc6 1 9 'iVa3 .!Dcb4. Nikolic-Hulak, Zagreb
1 987, continued 20 .:t c4? .!Da6 2 1 .!Dd6 .!Dac7 22 .:t ac l .!DbS 23
.!DxbS ..txbS 24 : cS ..te8 2S b4 b6, and White achieved nothing;
but Gutman's suggestion 20 .!Dc3! gives him the advantage. At
move 1 7, an alternative is 1 7
..tc6 18 .!DeS .!Dc7 19 ..txc6! with
the initiative; this is stronger than 18 .:t fc l a6 ( 1 8 . . . .!Dc7! at once
is more precise) 19 .:t c4 ( 1 9 .!DeS! is correct) 19 . . . .!Dc7 20 .!Dc3
..txg2 21 xg2 .!bedS, with equality; Andersson-Hulak, Wijk
aan Zee 1 987.
14 . . . 'iVxd7 15 e5 .:l.fc8 16 :t el ..t/8 17 ..tj3 .:l. c7
An accurate move which equalises. I would have answered the
incautious 1 7 . . . a6 with 1 8 ..te2 .:t c7 1 9 g4 ! ..th6 20 'iVd2, and
if 20 . . . .!Dc4, then 2 1 ..txc4 : xc4 22 .!De4 ! . But now this idea
doesn't work : 18 ..te2 : ac8 19 g4 ..th6 20 'iVd2 .!Dc4 21 ..txc4
.:t xc4, and both black rooks are on the c-file.
18 b3 : ac8 19 'iVd2 lbc6 20 'iVb2 a6 21 ..te2 'iVe7! 22 l1Jb1 liJb4
23 lbc3 lbc6 24 l:lJb1 lbb4 25 .:l. c5 t:tJd7 26 : xc7 : xc7 27 lbc3
l:lJ c6 28 l:lJb1 t:tJb4 29 l:lJc3 lbc6 30 lbb1 i-J.
In the third game, Kasparov varied with 1 2 . . . CfJe7! (a refinement on
12 . . . ..td7 of the first game) 13 a4 aS 14 'iVb3 (Black would have
had more problems after 14 e4 de l S ..txe4 .!DbdS 16 'iVb3.
Greenfeld-Birnboim, Tel-Aviv 1 988, continued 16 . . . f6 17 .!Dc4
h8 1 8 .:t fe l CfJb4 1 9 .:t ad 1 .!bedS 20 .!De3 : a6 2 1 ..txdS ed 22
.!DexdS ..tg4 23 .:t d2 .:t e6 24 .!Dxb4 .:t xe l + 2S ..txe l ab 26 'iVxb4
: e8 27 ..tf2, and White gained the advantage) 14 . . . ..td7 1 S
.:t fc 1 ..tc6 1 6 .!DbS .!Dbc8! 1 7 e 3 .!Dd6 1 8 .!Dxd6 'iVxd6 1 9 .tel
.:t fb8 20 ..tfl f6 2 1 t:tJf3 'iVd7 22 'iVc2 .!DfS 23 ..td2 .!Dd6 24 b3
.:t c8 2S 'iV d l h6 26 .te l gS 27 : a2 'iVe8 28 : ac2 .U8 29 ..td3
g4 !-!.
. . .
12
b3
This move was played for the first time in Portisch-Korchnoi,

Reggio Emilia 1 987/8. It seems to me that if White is seeking the
initiative, it is worth trying 1 2 .!Dxc6 be 1 3 ..tf2.
12
13
..td7
f6
'iV d2
In Portisch-Korchnoi, Black first played 1 3 . . . .:t e8 1 4 .:t fc l ,
and only then 1 4 . . . f6. After l S .!D d 3 .:t e7 1 6 h 1 ..te8, the
players agreed a draw. Evidently on that day they were just not
in the mood for a fight; the result has nothing to do with the
position. In the present game, I carry out a regrouping in the
centre as Black, without wasting time on rook manoeuvres.
14
15
16
IL!d3
.l:t acl
ILleS
IL!d6
-*.f2 (1 12)
JJ2
B
The position is characterised by so-called dynamic equilibrium.

The pawn chains are closed, and the forces cannot clash for some
time yet. Some precise and none-too-obvious manipulation of the
pieces is called for, and of course such operations are not to
everyone's taste. But I personally have always liked this kind of
game, based as it is on a wealth of subtle points. I had previously
encountered such situations with White, but was also ready to try
them with Black, especially since the configuration of pieces is
nearly symmetrical.
Here, by the way, is an illustration of the subtleties inherent in
the position. Black just needs to play one careless move, 16 . . .
..te8?, and White immediately seizes the initiative : 1 7 ILleS! -*.f7
( 1 7 . . . "W/e7 18 IL!xe6!) 18 e4 de 19 ILJ3xe4 IL!xe4 20 -*.xe4 "W/e7 21
IL!xb7!
16
17
18
19
IL!e5
h1
g4!?
f5
"W/e7
.l:t fc8
1 9 IL!a4 or 19 a3 would have preserved equality, but White is

trying for more.
19
20
21
IL!xe4
IL!xd7
IL!e4
fe
The knight cannot hold out on eS for long, but perhaps it was
worth exchanging it for the black one : 2 1 .l:. c3 xeS 22 de .l:. xc3
23 'ii' x c3 .l:. c8 24 'ii' d 2 b6 2S .tg3, intending f4-fS.
2 1 .tg3!? at once is also interesting : 2 1 . . . xeS 22 de .l:. xc l
23 .l:. xc l .l:. c8 24 .l:. xc8+ .txc8 2 S 'ii' c 3 (White's aim i s t o carry
out f4-fS, but the situation does not yet call for it : 2S fS? ef 26 gf
gf 27 'ii' xdS + ..te6, or 2S 'ii' d 4 b6 26 fS ef 27 gf gf 28 'ii' xdS +
J. e 6 2 9 'ii' a8 + f7, with a good game for Black i n either case)
2S . . . 'ii' d 7 (but not 2S . . . 'ii' d 8 26 .th4!, or 2S . . . .td7 26 'ii' c 7)
26 e3 b6 27 a4 .tb7, and again the chances are about equal.
21
22
'ii' xd7
a3 (1 13)
l /3
B
One thoughtless move, made on general grounds - and already

Black has a chance to seize the initiative. 22 e3 was correct,
avoiding weakening the queenside.
22
23
.th3
aS!
'ii' e7
Forcing the rook to abandon the c-file.

l4
.l:. al
d8
Black could have prevented the f4-fS break by playing 24 . . .

.th6 2S gS .tg7. After 26 .l:. fcl, followed by e2-e3 and .th3-fl ,
it would be hard for either side to play for a win. Instead, I attempt
to bring my knight over to the kingside, where the decisive events
will shortly take place.
25
fS
gS
The sharp 2S . . . f7!? is also interesting.

26
.tg3
c6
Fianchetto System 1' 77

At this point I convinced myself that after 26 . . . lll f 7 27 fe'!
1Wxe6 28 : f5 h6 29 : an lll d 6 30 ..te5! lll xf5 3 1 gf 1Wd7 32 ..txg7
1Wxg7 33 f6 the initiative is with White, so I decided to bring the
knight back again.
27
28
29
30
e3
fe
.:. rs
1We2
.:t d8
1Wxe6
h6
30 .:. afl looks more consistent, yet White cannot successfully

exploit his domination of the f-file : 30 . . . lll e 7 3 1 ..tc7 .l: dc8! (but
not 30 . . . .l: e8 3 1 ..te5 lll xf5 32 gf 1Wa6 33 ..txg7 xg7 34 f6 +,
with an attack for the exchange), and now 32 ..te5 is not so
effective : 32 . . . lll xf5 33 gf 1Wa6! 34 ..txg7 (34 f5 .:t f8!) 34 . . . xg7
35 f6 + h8! 36 1Wf2 .:t c7 37 ..td7 .l: d6 38 f7 .:t f8 39 1Wf5 g7
40 1We5 + .:t f6 4 1 ..tb5 1Wd6, and White's attack is spent. On the
other hand after 32 ..txa5 Black takes the exchange, and his
chances must be preferable.
30
31
32
33
34
.:. n
: act
.:t xc6
.:t bl
lll e7
.:t dc8
.:t c6
1Wxc6
Rather a passive move. It was worth considering 34 .:t f2!?, and

if 34
'ii' c l + 35 ..tfl 1Wxa3, then after 36 1Wb5 1W b4 (36 . . . a4?
37 1Wxb7 ab 38 ..td6) 37 1Wd7 White has a dangerous initiative
for the pawn. Instead, 34
a4! leads to unclear play : 35 ba (better
than 35 b4 1Wc1 + 36 'ii' fl 1Wxa3 37 ..td6 lll g6 38 1Wb5 'ii' c l + 39
..tfl 1Wc6 with advantage to Black, or 36 ..tfl 1Wxa3 37 1Wb5
1Wxe3 38 1Wxb7 .:t f8 etc.) 35 . . . .:t xa4 36 1W b2.
. .
. .
34
35
36
37
38
..tn
1W b2
.tel
.tel
..tf8
lll c8
1Wd7
..td6
The advantage of the bishop pair is illusory - it was better to

eliminate the bishop on d6. After 38 :n (more precise than 38
..txd6 lll xd6) 38 . . . ..txg3 39 hg, it would be difficult to breach
White's fortress.
38
39
.:. cl
lll e7
: f8!
40
g2
Not 40 J.xa5?, on account of 40 . . . l: f2. But 40 gl, followed

by J.e2-d l , was more tenacious.
ll:l g6
40
41
42
b4
.-c2
a4
.-f7!
The threats on the f-file are becoming formidable.

43
.-dl
The black a-pawn is immune : 43 .-xa4 ll:lf4 + ! 44 ef .-xf4 45

J.g3 .-xc l 46 J.xd6 .-e l l 47 J.xf8 .-xe2 + 48 gl e3 49 h3
.-f2+ 50 h l e2, and wins.
h8!
43
Preparing to open the g-file.

44
l: c3 (1 14)
44 g l would likewise be answered by 44 . . . h5!, intending to
take the g-pawn; after 45 gh .-f5! 46 hg (46 g2 ll:lh4 + 47 J.xh4
gh is crushing) 46 . . . .-h3 47 J.g3 J.xg3 48 hg .-xg3 + 49 h l
l: f2 5 0 l: c8 + g7, i t i s all over.
Nor does 44 l: c2 help : 44 . . . h5! 45 gh ll:le7 (45 . . . ll:lh4+ is
premature because of 46 J.xh4 gh 47 J.g4) 46 h3 (46 l: c3 g4!
and . . . .-xh5) 46 . . . ll:lf5 47 l: c3 ll:lh4+ 48 ..txh4 gh 49 .-el (49
.-gl l: g8 + 50 .tg4 .-r3 mate) 49 . . . .tg3 50 .-n .-f2+, and
Black wins.
1 14
B
44
45
46
gh
J.xh4
hS!
tt:Jh4 + !
I f 4 6 gl (46 hl .-fl +), then 4 6 . . . .-fs, followed b y . . .

.-f5-h3, is decisive.
46
47
48
et
gh
h3+ !
hl
Other continuations are less stubborn : 48 g1 g7 + , or 48

xh3 : g8 49 ..tg4 : xg4 50 xg4 B + 51 g5 ..te7 + 52
h6 f6 mate.
48
49
50
51
52
53
x2
: cl
hl
2
: xfl
: xhl+
: gl+
: fl
55 ..td3 is met by 53 . . . : f3.

53
54
55
56
57
58
59
60
61
hl
hl
hl
fl
:ci
..tg3
h7
h6
: hl +
: 2
..td6
: gl +
: g8
: g3
6 1 . . . ..th2 was immediately decisive; so was 6 1 . . . ..txb4 62

ab h2 63 f2 : g 1 64 : xg 1 hg()+ 65 xg l a3.
The game was now adjourned. Realising that Black's threats
were unanswerable, White resigned without playing on.
0-1
Game No. 25
Vaganian-Kasparov
Barcelona 1989
1
2
3
4
5
6
103
c4
g3
..tg2
10c3
cd
106
g6
..tg7
0-0
d5
10xd5
The reader will recall that in the World Championship matches

of 1 986-7, Kasparov preferred the symmetrical system, in which
Black prepares . . . d7-d5 with . . . c7-c6 and recaptures on d5 with

the pawn. In a game we played later (and which I shall examine
further on), he chose . . . <'i)xd5. In the last two or three years,
Kasparov has played a number of quite interesting games with
this line, and the one I have selected here as the 'principal' game
is among the most fascinating of all. In the notes, as usual, I shall
also mention other important examples from recent practice.
7
0-0
<'i) c6
8
d4 <'i)b6 (1 15)
1 15
w
Here White has the choice between the slow e2-e3, as in the
actual game, and the more committal d4-d5 and e2-e4. Let us
first consider the latter.
9 d5
Another possibility, seen more rarely, is 9 ..tf4.
9 . li)a5 10 e4
The modest 10 'ii' c 2 quickly leads to equality; here are two
examples :
Tukmakov-Halifman, Simferopol 1 988, went 10 . . . c6 1 1 de
<'i) xc6 12 .l:l. d1 ..tf5 1 3 e4 ..td7 14 -*.f4 .l:l. c8 1 5 'ii' e 2 'ii' e 8 1 6 h3
-*.e6 17 <'i)d5 ..txd5 18 ed <'i)b4 19 <'i)e1 'ii' d 7 20 'ii' d2 <'i)a6 !-! .
Plachetka-Smejkal, Trnava 1 989, varied with 1 2 . . . 'ii' c7 1 3
<'i)b5 'ii' b8 1 4 ..tf4 e 5 1 5 ..te3 ..tf5 1 6 'ii' c5 .l:l. c8 1 7 <'i)xa7 <'i)a4 1 8
<'i)xc6 be 1 9 'ii' e 7 t-t.
In Nikolic-Kasparov, Skelleftea 1 989, White played instead 1 0
..tf4, and there followed 1 0 . . . c6 1 1 de 'ii' xd 1 1 2 .l:l. axd 1 <'i)xc6
1 3 <'i)b5 ..tg4! 14 b3 e5! 1 5 ..te3 e4! 1 6 <'i)fd4 lDxd4 1 7 <'i)xd4 <'i)d5
1 8 h3 <'i)xe3 19 fe ..td7. Black's position is the more pleasant,
though the game was eventually drawn.
.

10 . . . c6 11
i. g5
An inadequate alternative is 1 1 .J:I. e 1 tL! ac4 ! 1 2 1We2 i.g4 l 3 de

( 1 3 ..tg5 h6) l 3 . . . be 14 h3 ..t xf3 1 5 i.xf3; Yurtayev-Belov,
Podolsk 1 989. At this point, 1 5 . . . 1Wd6 (instead of 1 5 . . . aS), with
the idea of 16 . . . 1W c5 or 16 . . . 1W b4, would have given Black a
good game.
11 . . . h6
Black failed to j ustify 1 1 . . . i.g4 in Hansen-Kasparov, Thessa
loniki 01 1 98 8 : 1 2 h 3 ..txf3 l 3 1W xf3 cd 14 tL!xd5 tL!xdS 1 5 .J:I. ad 1
1W c7 1 6 ed ..txb2 ( 1 6 . . . tL!c4 1 7 1We2 .J:I. fe8 1 8 .J:I. c 1 b 5 1 9 b3 "ii e 5
20 "ii xe5 tL!xe5 21 d6, and Black is in a bad way) 17 .J:I. fe 1 .J:I. fe8
18 l: xe7! .J:I. xe7 1 9 d6 .J:I. e 1 + 20 .J:I. xe 1 "ii xd6 21 .J:I. e7 .J:I. f8 22 "ii d5,
and although Black did manage to save himself, his position does
not inspire much confidence.
12
i./4
cd 13 ed
13 tL!xd5 is strongly answered by 13
tL! ac 4 ! . In Pastircak
Hort, Czechoslovakia 1 984, White continued with t h e straight
14 tLJc7?, and ended up in a difficult position after 14 . . .

.1:1. b8 1 5 "ii xd8 .J:I. xd8 1 6 .J:I. ad 1 ( 1 6 tL!a6 e 5 ! ) 1 6 . . . ..tg4 1 7 tL!d5
e5 18 tL! xb6 .J:I. xd 1 19 .J:I. xd 1 tL! x b2 ! 20 .J:I. d2 ef. Instead 14 1W b3
forward
might appear dangerous for Black, but with 1 4 . . . e 5 ! 1 5 tL! xb 6

"ii xb6! 1 6 ..t x h 6 ( 1 6 "ii xc4 e f ) 1 6 . . . i. x h 6 1 7 1W xc4 1W xb2 1 8
.J:I. ab 1 1W a 3 1 9 tL!xe5 i.e6, he obtained the bishop pair and plenty
of chances for the pawn in Shpilker-Krasenkov, Moscow 1 987.
13 . . . tL!ac4 14 1We2 g5
If 1 4 . . . tL!xb2, then 1 5 tL!e5 ! .
15
i. c l
e6! 16 de
Better 1 6 .1:1. d 1 , although 1 6 . . . ed 1 'l tL! xd5 tL!xd5 1 8 "ii xc4 tLJ b6!
1 9 "ii b3 W f6 gives Black slightly the better game (Popovic).
16 . . . ..txe6 1 7 .J:I. d1 1W c8 18 tL! d4

A sounder line is 1 8 h4 ..tg4 1 9 hg.
18 . . . i.g4! 19 i./3 ..th3! 20 tL!d5 tL!xd5 21 ..txd5

23 ..txc4 W/5 24 ..td3 W e5 25 1Wxe5 ..txe5
.J:I. d8
22
tL! c6 be
Black has a clear endgame advantage, and went on to win in

Hjartarson-Popovic, Belgrade 1 989.
9
9
. . .
e3
e5
.J:I. e8 is also seen. Kasparov, for example, played it twice
10 d 5 tL!a5 1 1
tL!d4 i.d7. I n Saloniki, Portisch played 1 2 "ii e 2, which led to
against Portisch in 1 98 8 . Both games continued
equality after 1 2 . . . c6 1 3 de tL! xc6 1 4 tL! xc6 i. xc6 1 5 ..t xc6 be

1 6 l:l. d 1 'ii' c 8 1 7 .i.d2 'ii' e 6 1 8 .i.e1 aS. In Reykjavik, he innovated
with 12 b4?!, but after 12 . . . ac4 1 3 a4 aS 14 bS 'ii' c 8 1 S l:l. e 1
.i.h3 1 6 .i. h 1 'ii' g4 1 7 ce2 l:l. ad8 1 8 'ii' b 3 'ii' d 7 1 9 f4 .i.g4 20
l:l. a2 'ii' d 6, Black had some initiative in the complexities. All the
same, this game too ended in a draw.
In Ljubojevic-Kasparov, Barcelona 1 989, White introduced a
different novelty, though he moved the same pawn : 12 b3. However,
after five more moves - 1 2 . . . cS 1 3 de xc6 14 xc6 .i.xc6 1 S
.i.xc6 'ii' x d 1 1 6 l:txd 1 be 1 7 .i.d2 a S - a draw, once again, was
agreed.
Finally, another game by Kasparov went 12 e4 cS 1 3 de lt:)xc6
14 ltlxc6 .i.xc6 1 S 'ii' b 3 'ii' d 3 16 .i.e3 : ac8 17 l:l. fd 1 'ii' a 6 18 .i.d4
'ii' a S, with equality; Hjartarson-Kasparov, Reykjavik 1 989.
as
10
dS
Practice has shown that retreating with 10 . . . e7 is too passive.
11
12
e4
c6
.i.gS
At this point we shall digress to examine the game Karpov

Kasparov, Amsterdam 1 988, in which the same position arose,
only with the rooks on the e-file instead of the f-file. The order of
the opening moves was 1 d4 f6 2 c4 g6 3 f3 .i.g7 4 g3 dS
S cd xdS 6 .i.g2 b6 7 c3 c6 8 e3 0-0 9 0-0 l:l. e8 1 0 : e l
e S 1 1 d S aS 1 2 e4 c6 1 3 .i.gS (1 16) .
116
B
Play continued 1 3 . . . f6 1 4 .i.e3 ac4 ! 1 S de xe3 (Black is

willing to give up a small amount of material rather than accept
defects in his pawn structure. At the same time he activates his
forces) 16 'ii' x d8 : xd8 17 cb .i.xb7 (if 1 7 . . . xg2, then 1 8 ba('ii' )
xe1 1 9 'ii' x a7 destroys all Black's illusions) 1 8 : xe3 .i.h6 1 9

: ee l lt)c4 20 : ad 1 f8 2 1 h4 : ac8 22 .th3 (White appears to
have seized the initiative, but this is purely temporary) 22 . . . : xd l
2 3 : xd 1 lt)xb2 24 : d7 : xc3 2 5 : xb7 (it was worth testing Black
by first playing 25 : d8 + e7 26 : d7 + f8; the point is that
26 . . . we8 is risky, since the king needs to guard g7 as a retreat
square for the bishop) 25 . . . lt)c4 (the opposite bishop ending after
25 . . . : xf3 26 : xb2 is unpleasant for Black) 26 lt)h2 (not 26
: c7?, in view of 26 . . . : c 1 + 27 .tfl lt)d2 28 : xc 1 lt)xf3+ 29
g2 lt)xh4+ 30 gh .txc 1 , and now it is White who has to hold
an opposite bishop ending a pawn down. On the other hand, a
draw results from 26 .te6 lt)d6) 26 . . . lt)d6 27 : xh7 (or 27 : xa7
: xg3 + ! 28 fg .te3 + with equality) 27 . . . .tg7 28 h5 gh 29 .C. xh5
ll c l + 30 g2 : c2 31 .te6 lt)xe4 32 lt)g4 : d2 33 .tb3 (the last
chance was 33 f3 lt)g5 + 34 e3 : b2 35 .tb3, but 33 . . . lt)c5
34 .tc4 maintains the balance) 33 . . . a5 34 : f5 lt)d6 35 ll h5 lt)e4
36 : f5 lt)d6 37 : h 5 lt)e4 t-t .
12
13
14
.te3
ed
f6
cd
After 14 .txb6 'it'xb6 1 5 lt)xd5 'it'd8 1 6 b4 lt)c6, the chances

are approximately eq u al.
14
.tg4
Black's plans involve the advance . . . f6-f5, which may be carried

out with the bishop on either g4 or c8.
Mikhalchishin-Gavrikov, Budapest 1 989, went 14 . . . : f7 1 5
.tc5 f5 1 6 'it'cl ! ( 1 6 lt)d2 e4 1 7 lt)b3 li)ac4 1 8 'it'e2 lt)e5 1 9 : ad 1
lt) bc4 2 0 lt)d4 b6 2 1 lt)c6 'it'e8 led t o equality i n Gligoric-Gavrikov,
Moscow 1 989) 16 . . . e4 1 7 lt)g5 : c7 1 8 b4 lt)bc4 (other continu
ations are worse, for example 1 8 . . . lt)xd5 19 lt)xd5 'it' xd5 20 .C. d 1 ,
with a won position) 19 ll d 1 .txc3 20 'it' xc3 'it' xg5 2 1 .te3!
(peace was quickly concluded in Mikhalchishin-Ftacnik, Palma
de Mallorca 1 989 : 2 1 .td4 .td7 22 ba 'it'e7! 23 .tb6 : cc8 24
.tc7 lt)d6 25 .t xd6 'it' xd6 26 'it' d4 t - 1 . On the other hand, 2 1
'it'd4? b 6 22 d 6 : xc5 2 3 be .te6 i s good for Black) 2 1 . . . 'it'd8!
(better than 21 . . . lbxe3 22 'it'xc7 lbxd 1 23 : xd 1 ) 22 ba lbxe3 23
'it'xe3 'it'd6 24 f3 ef 2 5 .txf3 .td7 26 : ac t (better 26 : ab l . Now
the game is level) 26 . . . : xc 1 27 : xc 1 b6 28 'it'f4 'it'xf4 29 gf : c8
30 : xeS+ .txc8 3 1 ab ab 32 f2 f7 33 e3 we7 34 d4
d6 t- t .
15
.tc5

Practice has also seen 1 S h3 .txf3 1 6 .txf3 fS 1 7 .te2, with
complex play.
15
16
17
b3
l U7
fS
.tb4 (11 7)
117
B
A well-known position. Black normally used to play 1 7 . . . 'l:lc8,

whereupon White has the strong move 18 'lr e l . This occurred,
incidentally, in Vaganian-Thorsteins, Copenhagen 1 988. The idea
is that Black no longer has the important tempo-gaining move . . .
1rd8-b6 (it would be met by 'l:lc3-a4), his knight on aS is indirectly
attacked, and the position is very difficult for him. After 1 8 . . .
.txf3 1 9 .txf3 'l:ld6 20 .te2 b6 (better 20 . . . e4) 2 1 .ta6, White
gained the advantage and won.
But Kasparov has a remarkable surprise in store for his
opponent.
17
'l:lac4!
An unexpected piece sacrifice, although admittedly the material
is quickly regained. Kasparov had, of course, prepared this novelty
in his pre-game analysis, knowing that Vaganian regularly plays
this variation.
18
be
e4
White's pawn-couple looks strong, but Black has ample

resources. A notable point is that as long as the black knight was
on aS, the white bishop on b4 was splendidly placed, both attacking
and defending; but once the knight has sacrificed itself, the bishop
is constantly exposed to the threat of . . . a7-aS. Vaganian himself
commented on this paradox of the instant transformation of this
bishop.

19
J: c l
I f 1 9 cS, Black has the very strong counter-blow 1 9 . . . a S !
1i'd7?!
19
The situation is extremely tense, and Black fails to choose the

correct path. Evidently Kasparov underestimated his opponent's
retort. After the game, he demonstrated the following variations :
19
or
.!Oxc4 20 .!O xe4 fe 2 1 J: xc4 .txf3 22 1i' d 2 11' d7, with equality;
19 . . . ef 20 .txf3 .txf3 2 1 11' xf3 2 .!O xc4. The latter, however,
was played i n Ionov-Urban, Katowice 1 99 1 , and after 22 lObS

.!OeS 23 1i'b3 aS 24 .td6 a4 25 "ife3 f4 26 gf .!Og4 27 11'e6 11' h4 28 f5
1i' gS 29 fg .!O e S + 30 h l .!O xg6 3 1 J: c7 White neutralised his
opponent's initiative and went on to win.
20
21
22
23
24
.!Ob1!
.txf3
11'xf3
g4
1i'b3
ef
.txf3
f4
11'a4
This time it is White who plays inaccurately. After 24 .tc3!

.txc3 2 S .!O xc3 11' xc4 26 J: fd 1 J: d8 27 d6, he could feel fairly
confident.
1i' d 7
24
The players are committing errors alternately. Black sho uld

have played 24 . . . f3, with a good game for the pawn; White's
king is badly placed, and his pieces are rather passive.
25
f3
h5 ( 1 18)
26
27
h3
1i'd3
J: e8
/ 18
w

It was not worth chasing after a second pawn. After 27 .i.c5!
'IJJ/ c 7 28 .i.f2 .l::t e3 29 'IJJ/ d l , White would have set up a solid
protective wall while keeping a material advantage. Now the
mutual turmoil subsides, and the game quickly heads towards a
peaceful result.
27
bg l8 bg .l::t e3! 29 'IJJ/ xg6 .l::t f6 30 'IJJ/ gS 'IJJ/ f7 31 .tc3 .l::t g6
32 'IJJ/ d8+ .l::t e8 33 'IJJ/ h4 .i.xc3 34 .l::t xc3 'IJJ/ g7 35 .!::t el 'IJJ/ d4+ 36
2 .l::t xel 37 'IJJ/ x el xc4 38 'IJJ/ e8 + g7 39 'IJJ/ e7 + b6 40
'IJJ/ h4 + g7 41 'IJJ/ e7 + b6 42 'IJJ/ fB + b7 43 'IJJ/ f7 + b6 t- t

Index of Variations
1
d4 lL!f6 2 c4 g6
Seville Variation
3
lL!c3
4
cd
5
e4
6
be
7
.tc4
8
lL!e2
9
.te3
10 0-0
lO . . . ed 1 1
f3
II
12
.txf7 +
13
fg
14
xfl
1 4 . . . ed
d5
lL!xd5
lL!xc3
.tg7
c5
lL!c6
o-o
.tg4
lL!a5
.l:txf7
l:t xfl +
'ii' d6
ed 'ife8 15
. . 'ii' d 7 34
1 5 . . 'ii' b6 35
1 5 . . . e5 1 6 de 3 7
1 6 .:t e l 37
16 d 5 37
1 4 . . . 'Wd7 1 5 de 13
1 5 g5 'ii' e 6 13
1 5 . . . .:t d8 1 4
15
15
1 5 h3 1 4
14
. . .
1Ve8 15
15
15
e5
28
g 1 .:t d8 29
1 5 'if a4
188 1 ndex of Variations

1 5 . . . 'it'e6 16 'it'd3 cd 1 7 cd J:l. d8 1 8 g5 29
1 8 h3 30
16 . . . 'it'c4 17 'it'd2 30
1 7 'it'xc4 30
15
'it'd5
1 5 . . . 'it'e6 15
16
..tf2
1 6 ltlf4 16
J:l. 8
16
1 6 . . . J:l. d8 1 7 ltlf4 19
17 1i' e 1 19
17 'it' a4 ltlc6 20
1 7 . . . ltlc4 1 8 ltlf4 20
1 8 g5 20
1 1 . . . : r8 21
17 . . . b6 18 "ifc2 J:l. c8 21
18 . . . 'it' c4 22
1 8 . . . : rs 1 9 ..tg1 22
1 7 'it'c2 J:l. c8 25
1 7 . . . 'it'c4 1 8 g5 25
1 8 1i' e4 25
1 8 'it'b2 25
17
g5
1 7 g1 16
'it'f7
17
1 7 . . . 'it'e4 1 7
18
'it'el
h6
1 8 . . . 'it'f5 1 7
1 8 . . . ltlc4 1 7
19
gh
18
Exchange Variation with 7 ..tc4 - other systems

d5
3
ltlc3
4
cd
ltlxd5
5
e4
ltlxc3
..tg7
6
be
7
..tc4
c5
ltlc6
ltle2
8
9
..te3
0-0
10
o-o
Index of Variations 189

1 0 .l:[ b 1 46
1 0 : c 1 ..t d 7 46
10 . . . cd 1 1 cd 'if a 5 + 1 2 fl .l:[ d8
1 2 . . . 'if a 3
47
47
1 2 . . ..td7 1 3 h4 h 5 48
.
1 3 . . . e5 48
1 3 . . . : ac8 48
13 . . .
11
12
13
14
15
16
17
f3
..td3
cd
:cl
'if a4
d5
aS
cd
..te6
..txa2
..te6
..td7
: res
'ii' b4
1 7 'if a 3 b5 42
e6
17
1 7 . . . b5 42
1 7 . . . b6 42
18
: fdl
1 8 de 43
1 8 d 6 43
1 8 c3 43
1 8 d4 44
18
ed
44
Modern Exchange Variation

d5
c3
3
4
cd
xd5
5
e4
xc3
..tg7
6
be
6 . . . c5 83
7 ..te3 c5
f3
8 'ii' d 2 cd 84
8
. .
. 'if a 5 90
8 . . . 0-0 9 d5 9 1
9 : c 1 91
9 f3 c d 93
9 . 'if aS 1 0
. .
9 ...
l:t c 1 93
..t g4 1 0 g5 104
50

7
8
9
c5
0-0
cd
: bt
..te 2
9 . . . ..t g4
53
'ilf a5 1 0 l:t b5 53
1 0 0-0 'ilfxc3 53
10 . . . 'ilf xa2 1 1 ..tg5 53
9 . . . b6
1 0 ..te3 57
1 0 0-0 ..t b7 1 1 W' d 3 5 7
9 . . . tt:J c6 10 ..te3 61
1 0 d 5 61
10
'ilfa5+
cd
11
..t d2
1 1 'i!f d 2 67
11
W'xa2
12
0-0
..tg 4
1 2 . . . lZJ c6 68
1 2 . . . b6 1 3 'ilf c l ..tb7 68
9 ...
1 3 . . . 'ilf e6 1 4 ..tc4
69
. tt:Jd7 1 3 n a t 71
12
..
12
. . . W' e 6 1 3 W' c2
1 3 ..t b4
13
..tg5
71
77
79
Russian System
tt:Jc3
3
d5
4
tt:Jf3
..tg7
5
'ilfb3
de
6
'ilf xc4
0-0
7
e4
lZJa6
7 . . . ..tg4 8 tt:Jg5 108
8 tt:Je5 108
8 ..te3 tt:Jfd7 9
9
8
8 W' b 3
8 ..tg5
8 ..tf4
8
9
..te 2
1 15
1 15
1 15
c5
d5
: d l 109
W' b3 lZJb6
111
Index of Variations 191

9 e 5 1 16
9
de
1 16
9
10
10
.tg5
10
11
e6
0-0
116
ed
.t f5
ed
1 1 . . . b5 1 1 7
1 1 . . . b6 1 1 7
11
..
1 2 .l:t d 1 1 18
1 2 .tf4 b6 121
12 . . . .tf5 - see
.l:t e8
12
below
.tf4
1 2 l: d 1 l: e8 1 18
.tg5
1 2 .te3
12
132
132
: es
12
1 2 . . . 'i!V b6
1 2 . . . tt:Jd7
13
13
13
126
129
l:t adl
tt:Je4
. . . 1V b6 121
. . . tt:Jd7 121
14
.td3
14 tt:Jb5 121
14
15
.txc3
be
125
4 .t f4 System
3
4
tt:Jc3
.t f4
4 tt:Jf3 .tg7 5 .tf4 e5

4
5
6
e3
de
d5
6
de
.tg7
c5
6 .txb8 135
6
7
'i!Va5
: ct
7 cd 136
7 'i!V a4 136
135

7 tt:lf3 0-0 8 J:t c l de 9 .txc4 xc5 10 .tb3 a5 1 1 0-Q tt:lc6 see below
7
de
tt:le4 144
7 .
.txc4
8
0-0
9
tt:lf3
9 tt:le2 147
9
xc5
.tb3
10
1 0 tt:lb5 148
10
tt:lc6
a5 1 1 0-0 tt:la6 148
10
a5
0-:0
11
h3 138
12
. .
. . .
Fiancetto System
.tg7
tt:lf3
c6
g3
4
4 . . . d5 5 cd tt:lxd5 6 tb c3 0-0 7 0-0 tt:lc6 8 d4 tt:lb6 9 d5 180
9 e3 ll e8 181
e5 182
9
d5
.tg2
5
. . .
cd
6 tt:lc3 0-0 7 b3 152

6
cd
ltl c3
0-0
tt:le5
e6
8 . . . .tf5 153
0-0
9
9 .tg5 154
9
9 . . . c6 160
10
tt:lfd7
f4
1 0 tt:lf3 160
10
10 . f6 160
10 . tt:lxe5 161
tt:lc6
. .
. .
11
.te3
1 1 . . f6 164
.
tt:lb6 1 73
BFAfiNC HIE (,R0 NFH D

'
'Si nce his surprise choice of the GrOnfeld Defence in the 1 986
World Championship match, Gary Kasparov has regularly
adopted the GrOnfeld in tournaments and matches around
the world. Against Karpov, however, this opening has proved
an unfortunate choice a,nd Karpov has won more World
Championship games against Kasparov in this opening than
in any other. Highlights of this book i nclude:
Original winning ideas against the GrOnfeld

Unique i nsights i nto Karpov's World Championship
preparation
Detailed coverage of key fashionable variations ,,
Fascinating tips from a former World Champion
No a mbitious player can afford to take on or play the

GrOnfeld Defence without this book!
Anatoly Karpov is one of the g reatest p layers of all time.
During his time as World Champion between 1 975 and 1 985,
his tournament record was unprecedented as he outclassed
his contemporaries time and again. His fierce rivalry with Gary
Kasparov has now endured for five matches for the world title
and numerous tournaments around the world.
a me i n Action
:>en Game in
rUI U t;UI I l plete l iSt Of theSe and

other Botsford chess books please
write to:
B.T. Batsford Ltd

4 Fitzhardinge Street
London Wl H OAH
ISBN 0 -7 1
-'=-PUBLISHER'sJ
\
net
prlca
,l
111
9 7 8 0 7 1 3 "4 0 4 0 '1 "

Beating The Grünfeld PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Beating The Grünfeld PDF

Încărcat de

Drepturi de autor:

Formate disponibile

Beating

Beating the Griinfeld

Beating the Griinfeld

B.T. Batsford Ltd, London

First published 1992

A BATSFORD CHESS BOOK

Exchange Variation with 7 c4 - other systems

Modern Exchange Variation

endgame. In some games, the later stages are of independent

The game has followed the main system of the Griinfeld. I

So International Master Kuzmin, one of my seconds, has

Let me emphasise that the main feature of the position is not

gh + .l:l. xh7 33 11fg4+ h8 34 .l:l. f l 11ff5 (34 . . . 11rd3 35 -*.h4!) 35

one other option : 1 4

In the 1 1 th match game in Seville, I chose 15 "'gl . For the

So far, the play coincides with game 5 in Seville. In the 7th

Gutman gives the variation 1 7 . . . "ife4 (or 1 7 . . . tt:Jc4 1 8 g1

lt:lxf2 (Kuzmin recommends 22 . . . "it'c4 + , without the exchange

Stronger than 19 lt:lg3 hg 20 lt:le4 cd 21 cd "it'f5.

Otherwise White is a clear pawn up. The bishop move to d2

25 . . . "it'f7 or 25 . . . llf4 would be safer.

If 26 "it'g4 (26 lt:le4 ..td4!), Black has 26 . . . llxf2! when Henkin's

After 26 . . . "it'e6 or 26 . . . "it'f7, Black could still offer resistance,

Now after the reply 1 7 . . . b6, my queen would withdraw to c2.

In the event of 17 .. . lLlc6 1 8 "it'b3 c4 19 'iltxb7, White has the

After this, the black pieces come under pressure. Evidently 18

From diagram 7, here are some important recent examples :

In Douven-Ilincic, Alma-Ata 1 989, after 1 9 1W d 1 l:t d8, White

White continues with his plan of restricting the bishop's mobility.

20 . . . 1Wf7 is not good, since after 21 .ltlg3 everything fits together

A critical moment. In addition to .!tle2-f4, White will now be

22 . . . -tf6 is no good, if only because of 23 ef (23 .ltlf4 is also

(for example) 24 ll:lf4 'flf7 25 ll:ld3 b4 !? even though 25 .tg3

This move was roundly condemned by the commentators. But

Forced. If 27 . . . b4, then 28 ll:lf5 is unpleasant, while 28 . . . .l:l. f7

29 . . . be loses immediately to 30 o!Llf5 <i>f8 3 1 e6 .l:l. xf5 32 gf

30 o!Llf5 is also strong: 30 . . . e6 3 1 o!Llxh6 .l:l. f4 32 1W xg5 , or 32

The exchange sacrifice is incapable of improving matters. All it

Black's pieces on the kingside never succeeded in breaking free.

But not 1 8 g5? on account of 1 8 . . . cd. If 1 8 'ilt'e4, then 1 8 . . .

This position first occurred in Karpov-Kasparov, Amsterdam

The stock manoeuvre; here too it guarantees White a plus.

Gaining a couple of tempi for bringing the queen to the defence.

But not 23 lbc l ? cd and everything is fine for Black, as after 24

25 lLif5 is also strong : 25 .. . 1t'xg5 26 lLixg7 irg4 (26 . . . xg7

An exchange sacrifice (again not entirely voluntary!) analogous

On 27 . . . lLic6, I had in mind 28 cd lbxd4 29 1i'e3 lbc6 30 1t'b3+

. tl:lxe6, then 34 l:t fl ! is decisive.

There is no defence against the threats of 35 tl:lf6+ and 35 : fl.

In the 5th and 7th match games in Seville I continued 1 5 e5,

A bold piece sacrifice. Instead, the timid 1 5 . . . b6 1 6 g 1 cd

20 . . . ba 2 1 lllf4 'flf7 22 lll d 5 .l: f8 23 i.g3 lllb 6 (but not 23 . . .

Seirawan-Lputian, St John 1 988, continued 18

Black can play 1 8 . . . eS at once, without opening the c-file.

After 19 . . . b6 or 19 . . . bS, White plays 20 :lbl! with the better

20 :lcl d6 2 1 de .i.xeS is not dangerous for Black.

A refinement on the 1 1 th game in Seville. But before going any

part in the match) proceeded. It might, incidentally, have been

White proceeds ineffectively and is soon in a difficult position.

White's loss of time in moving his bishop about begins to tell:

So 1 2 J.xf7 + doesn't guarantee victory. One or two inaccur

Curiously enough, this precise position arose as long ago as

The only move. There would be a striking finish after 25 . . .