Attention-deficit hyperactivity disorder (ADHD) is a prevalent childhood disorder that is often maintained throughout development and persists into adulthood, leading to substantial problems in daily functioning. Although pharmacotherapy is considered the first-line treatment for ADHD at all ages (at least for severe cases), non-pharmacological therapies might be equally effective without the risk of drug side-effects. Some studies have shown that electroencephalographic (EEG) neurofeedback improves parent-rated ADHD symptoms in children and adolescents. However, adequate and well controlled studies to support its use are scarce, particularly in adults. We searched PubMed with the search terms “ADHD”, “neurofeedback”, “adults,” and “clinical trial” for articles published in English between Oct 1, 2010, and March 31, 2017. This search identified two feasibility studies that investigated the effects of neurofeedback when compared with either no intervention or a sham control in small groups of students prone to ADHD. Although the first study of neurofeedback versus no intervention reported beneficial effects, the second study of neurofeedback versus sham control found no advantage of neurofeedback. The PubMed search additionally identified one randomized controlled trial. This trial showed that, compared with waiting-list control, neurofeedback promoted improvement of self-reported ADHD symptoms but did not show transfer of learning to a computerized assessment of executive functions. Added value of this study In this first randomized, adequately controlled, triple-blind investigation of neurofeedback for the treatment of adults with ADHD, we found no superior efficacy of neurofeedback to sham neurofeedback or meta-cognitive therapy in the treatment of ADHD symptoms. Furthermore, targeted EEG bands (θ:β) remained unaffected by neurofeedback training, and clinical improvements were unrelated to the EEG θ-to-β power ratios across assessments in all treatment conditions. Implications of all the available evidence This study adds to first evidence from other studies that investigated the effects of neurofeedback in children with ADHD or other clinical disorders and observed no advantage for neurofeedback when compared with sham treatments. Our results suggest that although neurofeedback training is effective in reducing ADHD symptoms it neither outperforms sham neurofeedback nor group psychotherapy. As such, neurofeedback cannot be recommended as an efficient approach in the treatment of adults with ADHD. 2- sessions. In the sham neurofeedback group, 15 sessions of sham neurofeedback were followed by 15 sessions of verum neurofeedback over 15 weeks. Participants in the meta-cognitive therapy group received 12 weekly scheduled group sessions of CBT-oriented psychotherapy. We decided not to provide a pure sham condition over 30 sessions on the basis of ethical considerations assuming that participants would benefit only from real but not from sham neurofeedback therapy. Thus, by administering proper neurofeedback to all participants, we ensured that everyone eventually received actual treatment.13 Neurofeedback training was done by experienced master-level clinical psychologists who were trained in neurofeedback and masked to the treatment condition. The meta-cognitive therapy was led by two clinical psychologists experienced in ADHD therapy for adults and trained in meta-cognitive therapy. A commercially available neurofeedback system (NeXus-32; MindMedia, Herten, Netherlands) and software (BioTrace+ version 2012) were used for neurofeedback and sham neurofeedback protocols. Neurofeedback training was implemented for EEG activity at the vertex, with use of a linked mastoid reference scheme and a sampling frequency of 256 Hz. Impedance was kept below 5 kΩ. The EEG signal was high-pass filtered at 1 Hz and low-pass filtered at 45 Hz. Every session consisted of a resting state (baseline) condition of 4 min with participants’ eyes closed, three training epochs (7 min each), and a transfer phase without feedback for an additional 7 min. The aim of the neurofeedback training was to decrease θ activity (4–8 Hz, as indicated by a histogram on the left side of the screen) while increasing β activity (13–21 Hz, represented as a histogram on the right side of the screen). Positive feedback was provided by visual effects (moving bubbles in the middle of the screen) when θ activity remained below threshold and β activity above threshold, otherwise bubbles stopped moving (appendix). The reward thresholds were automatically adjusted every 15 s (on the basis of real-time signal) to provide positive feedback about 80% of the time.15,16,22 The feedback process was automatic for standardization purposes, to minimize variability in terms of the type of feedback provided, and to enhance replicability of the procedure. The sham neurofeedback protocol was identical to the verum neurofeedback protocol except for the origin of the EEG signal displayed on the participants’ screens. In the sham condition, participants were presented with a pre-recorded signal from an individual who participated in the verum neurofeedback group of the study. Therefore, at the beginning of the training, a computer program allocated each participant of the sham neurofeedback condition to a participant of the verum neurofeedback condition and this assignment remained stable throughout the treatment. At the start of each neurofeedback session, the trainer entered the participant’s identification details and the computer program subsequently assigned the participant to the respective session signal of the matched verum participant (sham condition) or the verum condition (real signal). This allocation procedure was done while the monitor was black; therefore, the trainer was never aware of the participants’ randomization status throughout the therapy.