Research Brief · The Evidence, Plainly
The Mirror Talks Back
Neurofeedback has been sold for fifty years as two different things, a drug-free fix for attention, mood, and performance, or unproven theater dressed up as medicine. In 2025, JAMA Psychiatry published the largest and most rigorous test either side had faced, a meta-analysis of 38 randomized trials and 2,472 people, scored by raters who did not know who received the real training. Pooled across every trial design, neurofeedback showed no meaningful benefit for ADHD symptoms. Restricted to the three protocols with an actual clinical standard behind them, a small but real effect appeared, and the clearest single signal showed up on one cognitive measure, processing speed. This brief sets out what the blinded meta-analysis found, what an earlier review had claimed before that re-test, what a separate literature on mood and anxiety applications shows, how thin the adult evidence base still is, and what none of it proves.
What was actually studied
The largest and most rigorous test neurofeedback for ADHD has had. The authors pooled 38 parallel-arm randomized controlled trials, 2,472 participants aged 5 to 40, and focused the primary analysis on outcomes rated by people who probably or certainly did not know which group a participant was in. Pooled across every trial design, blinded ratings of total ADHD symptom severity showed no significant improvement with neurofeedback (standardized mean difference 0.04, not significant). Restricting the analysis to trials that used one of the three established standard protocols produced a small but statistically significant improvement (standardized mean difference 0.21). Of five neuropsychological outcomes examined across the full set of trials, only one showed a significant improvement: processing speed (standardized mean difference 0.35). The authors conclude that neurofeedback did not meaningfully benefit ADHD at the group level overall, and suggest future work focus on standard protocols, processing speed, and precision-medicine approaches to identify who might benefit.
The prior consensus this issue’s finding narrows. This review, co-authored by one of the field’s most-cited neurofeedback researchers, summarized meta-analyses and large multicenter trials available at the time and concluded that three standard protocols, theta beta ratio (TBR), sensorimotor rhythm (SMR), and slow cortical potential (SCP) training, were efficacious and specific for ADHD. It called for standardized practitioner training and binding clinical standards, on the premise that the modality’s evidence base was solid but its delivery in practice was not. The 2025 blinded meta-analysis above is a stronger test of the same three protocols and finds a real but smaller and narrower effect than this review’s framing implied, which is the gap this brief is built around.
Evidence that neurofeedback’s evidence base extends beyond ADHD, with the same pattern of a real but modest and specific effect. Eighty-seven patients with major depressive disorder and comorbid anxiety symptoms were assigned to alpha-asymmetry neurofeedback, high-beta down-training neurofeedback, or a control group, with ten sessions of training. Depression and anxiety scores decreased in both active neurofeedback groups relative to baseline, with no significant difference between the two active protocols. The high-beta protocol produced a measurable drop in high-beta power at the parietal cortex; the alpha-asymmetry protocol did not produce a significant change in its own target measure, the frontal alpha-asymmetry score, even though symptoms improved. That mismatch, symptoms moving without the targeted EEG measure moving, is a caution against assuming a protocol works for the mechanistic reason its name implies.
A mechanistic follow-up, not a treatment trial. Building on a separate real-time fMRI-and-EEG neurofeedback study in major depressive disorder, the authors applied source-localization analysis to the EEG data and found changes in prefrontal and amygdala current-source measures consistent with the frontal EEG asymmetry patterns the training targeted, with the prefrontal change correlating with reduced anhedonia severity and a separate measure correlating with reduced comorbid anxiety. This is exploratory, secondary analysis of a small existing dataset, useful for understanding a plausible mechanism and not a claim of clinical efficacy on its own.
The reason this brief is cautious about adult neurofeedback specifically. This review of interventions for ADHD in adults pooled 113 randomized trials and roughly 14,900 participants across pharmacological, psychological, and neurostimulatory or neurofeedback approaches. Neurostimulation and neurofeedback combined accounted for only ten of those trials and 194 participants, the smallest evidence base of the three categories by a wide margin. Stimulant medication and atomoxetine were the interventions with clear evidence of benefit on both self-reported and clinician-reported measures. The review does not report a neurofeedback-specific effect size for adults because the adult trial base is too thin to isolate one, which is itself the finding worth knowing before assuming the ADHD evidence above, drawn mostly from younger participants, transfers directly to an adult reader.
What they found
Pooled across every trial design and scored by raters who did not know who received real training, neurofeedback showed no significant benefit for ADHD symptoms as a whole. That null result is the headline the 2025 meta-analysis produced, and it is the honest answer to fifty years of marketing that promised more (Westwood et al., 2025).
The effect is not uniformly absent. Restricted to the nine trials that used one of the three long-standing standard protocols, a small but statistically significant improvement appeared. Restricted instead to specific neuropsychological measures across the full trial set, only one of five showed a significant gain: processing speed, with a similarly small effect size. The two restrictions are separate analyses of the same trial pool, not a single chain, and both point the same direction, toward a real effect that is narrower than the general ADHD-symptom claim most marketing makes (Westwood et al., 2025).
Before this blinded re-test, the standing narrative review in the field described the same three standard protocols as already efficacious and specific, based on the meta-analyses and multicenter trials available at the time, and called mainly for better practitioner standardization (Enriquez-Geppert et al., 2019). The 2025 meta-analysis is a stronger test of the same claim and finds a real effect, just a smaller and more specific one than that review’s framing implied.
Neurofeedback’s evidence base is not limited to ADHD. In a trial of 87 patients with major depressive disorder and comorbid anxiety, two different neurofeedback protocols, alpha-asymmetry training and high-beta down-training, both produced reductions in depression and anxiety scores relative to a control group (Wang et al., 2019).
In that same trial, the protocol named for its target, alpha-asymmetry training, moved symptoms without producing a significant change in its own targeted EEG measure, while the high-beta protocol did change its own target. Symptoms and mechanism did not move together in lockstep, which is a caution against assuming a protocol’s name explains why a patient feels better (Wang et al., 2019).
A separate, smaller mechanistic analysis of real-time fMRI-and-EEG neurofeedback in depression found prefrontal and amygdala source-activity changes consistent with the intended training target, with one change correlating with reduced anhedonia and another with reduced anxiety, offering a plausible account of mechanism without itself being a clinical efficacy trial (Zotev and Bodurka, 2020).
The adult evidence base is thin across the board. A 2025 review of ADHD interventions in adults, spanning 113 trials and nearly 14,900 participants, found only ten trials and 194 participants covering neurostimulation and neurofeedback combined, the smallest category by a wide margin next to medication and psychological therapies (Ostinelli et al., 2025).
Where the evidence stands
The well-supported claim is narrow: for ADHD, using one of three specific, decades-old standard protocols, a small but real effect on symptoms shows up once outcome raters are blinded, and the clearest single signal lands on one cognitive measure, processing speed, not the full symptom picture. That is a genuine finding from the largest trial pool ever assembled on this question, and it survives the strongest test available.
The under-supported claim is the one built into most of the modality’s marketing: that a proprietary protocol will meaningfully improve attention, mood, sleep, and performance all at once. The pre-2025 narrative review in this field made a similar overreach in the opposite direction, describing standard-protocol efficacy in terms a stronger blinded test has since narrowed. Neither the miracle framing nor the theater framing survives contact with the actual trial data intact, and the mood and anxiety literature (Wang et al.; Zotev and Bodurka) shows real effects that do not always move in lockstep with the mechanism a protocol’s name implies.
What follows for a reader is a change in how a claim gets evaluated, not a verdict on the modality as a whole. A provider who names a standard protocol, points to blinded evidence, and frames the likely benefit as small and specific is tracking what the research actually shows. A provider whose pitch is broad, unnamed, or measured only by client satisfaction has outrun the evidence, whatever else is true about their sincerity or their price.
What this does not prove
The strongest evidence here is about ADHD, mostly in children and adolescents. The 2025 meta-analysis spans ages 5 to 40, but adult-specific neurofeedback evidence is far thinner, as the 2025 adult ADHD review shows directly: only ten trials and 194 participants for neurostimulation and neurofeedback combined, against thousands of participants for medication and psychological therapies. An adult reader should treat the ADHD findings above as suggestive for their age group, not as adult-specific evidence.
A standardized mean difference of 0.21 is a small effect by conventional statistical benchmarks. Real and statistically significant are not the same thing as large or life-changing, and this brief does not claim otherwise.
The depression and anxiety findings (Wang et al.; Zotev and Bodurka) come from a different condition, a different age range was not reported in the sources reviewed here, and a much smaller evidence base than the ADHD meta-analysis. They should not be read as extending the ADHD effect sizes to mood applications, or vice versa.
The mechanistic analysis (Zotev and Bodurka, 2020) is exploratory, secondary analysis of a small existing dataset. It offers a plausible account of how a protocol might work. It is not a clinical trial and does not on its own establish that the protocol is effective.
Most neurofeedback delivered in ordinary clinical practice is not run inside a blinded randomized trial. The evidence graded in this brief describes what controlled research has found, not what any individual provider’s outcomes will look like, and a satisfied client is not the same kind of evidence as a blinded trial result.
None of this is medical advice, and none of it is a treatment recommendation. Whether neurofeedback is worth pursuing for a diagnosed condition is a decision for a licensed clinician who can weigh a specific case, not a newsletter. No supplement is recommended anywhere in this issue.
What it means for you
Ask which protocol a provider runs, by name, because the real effect in the largest trial to date belongs to three specific standard protocols, theta beta ratio, sensorimotor rhythm, and slow cortical potential training, not to neurofeedback as an undifferentiated category. Ask how any claimed result was measured, because the honest studies rated outcomes blind, and a clinic offering only satisfaction scores or its own before-and-after numbers is showing the version of the evidence that disappears once blinding is added. Weight a provider’s pitch to the measure the evidence actually supports: the clearest signal in the largest trial landed on one cognitive measure, processing speed, so a narrow, specific pitch tracks the research and a broad one, covering focus, mood, sleep, and performance from a single protocol, has outrun it. Expect the real dose, typically twenty to forty sessions of standard-protocol training, since that is the course length the trials that found something actually used. And track one retestable number, a working memory task, a reaction-time measure, or a symptom scale, taken before and after a course, rather than relying on how it felt, since a small, real effect will not always feel like something happened. The newsletter walks through those five moves. The research above is why they are the five, and where each one runs out.
Go to the source
- Tier 1 · Meta-analysis Westwood, S. J., Aggensteiner, P. M., Kaiser, A., Nagy, P., Donno, F., Merkl, D., Balia, C., Goujon, A., Bousquet, E., Capodiferro, A. M., Derks, L., Purper-Ouakil, D., Carucci, S., Holtmann, M., Brandeis, D., Cortese, S., & Sonuga-Barke, E. J. S. (2025). Neurofeedback for attention-deficit/hyperactivity disorder: A systematic review and meta-analysis. JAMA Psychiatry, 82(2), 118-129.
doi.org/10.1001/jamapsychiatry.2024.3702 - Tier 2 · Narrative review Enriquez-Geppert, S., Smit, D., Garcia Pimenta, M., & Arns, M. (2019). Neurofeedback as a treatment intervention in ADHD: Current evidence and practice. Current Psychiatry Reports, 21(6), 46.
doi.org/10.1007/s11920-019-1021-4 - Tier 1 · Randomized controlled trial Wang, S. Y., Lin, I. M., Fan, S. Y., Tsai, Y. C., Yen, C. F., Yeh, Y. C., Huang, M. F., Lee, Y., Chiu, N. M., Hung, C. F., Wang, P. W., & Liu, T. L. (2019). The effects of alpha asymmetry and high-beta down-training neurofeedback for patients with the major depressive disorder and anxiety symptoms. Journal of Affective Disorders, 257, 287-296.
doi.org/10.1016/j.jad.2019.07.026 - Tier 2 · Exploratory analysis Zotev, V., & Bodurka, J. (2020). Effects of simultaneous real-time fMRI and EEG neurofeedback in major depressive disorder evaluated with brain electromagnetic tomography. NeuroImage: Clinical, 28, 102459.
doi.org/10.1016/j.nicl.2020.102459 - Tier 1 · Systematic review and network meta-analysis Ostinelli, E. G., Schulze, M., Zangani, C., Farhat, L. C., Tomlinson, A., Del Giovane, C., Chamberlain, S. R., Philipsen, A., Young, S., Cowen, P. J., Bilbow, A., Cipriani, A., & Cortese, S. (2025). Comparative efficacy and acceptability of pharmacological, psychological, and neurostimulatory interventions for ADHD in adults: A systematic review and component network meta-analysis. The Lancet Psychiatry, 12(1), 32-43.
doi.org/10.1016/S2215-0366(24)00360-2
Tier 1 means peer-reviewed primary research or meta-analysis, the strongest evidence. Tier 2 means an expert framework or smaller study that traces to peer-reviewed work. We grade every source so you can see the weight behind each claim.
Pegasus Realm publishes Wellness Team Culture and the practice resources behind it.