This review pooled 38 randomized trials of neurofeedback for ADHD. Participants were aged 5 to 40 years, 2472 in all. Most trials trained brain-wave activity (EEG). Only 3 trials studied adults.
The key test used probably blinded ratings, meaning raters were likely unaware who got neurofeedback. Across 20 trials, these ratings showed no clear improvement in overall ADHD symptoms. Results for children alone were similar. Inattention and hyperactivity also showed no clear change. Training based on brain blood-flow scans (fMRI or fNIRS) also showed no benefit.
There were two possible exceptions. Trials using established standard protocols showed a small improvement, but the authors say it likely falls short of clinical value. Processing speed rose a little, the only one of 5 thinking-skill measures to improve, though the authors question its clinical value. In exploratory comparisons without blinded ratings, methylphenidate had a larger effect on symptoms than neurofeedback. Neurofeedback did not differ from other non-drug approaches, cognitive training and physical exercise.
Most trials had flaws that could skew results (high risk of bias), mainly poor blinding. The processing speed gain may be inflated, because it drew on linked estimates from 3 trials that shared comparison groups. Very few measured daily functioning, quality of life or school results. Too few allowed blinded analysis of adults or of effects beyond 3 months.
The authors conclude that the probably blinded evidence does not support neurofeedback as a stand-alone ADHD treatment at the group level.
The numbers Stats and what the terms mean
The statistics behind the plain-language claims above.
- Westwood, 2025: overall ADHD symptoms on probably blinded ratings, all trials, k = 20; n = 1214; SMD, 0.04; 95% CI, −0.10 to 0.18 (P = .56; I2 = 31%).
- Westwood, 2025: overall ADHD symptoms on probably blinded ratings, standard protocols only, k = 9; n = 681; SMD, 0.21; 95% CI, 0.02 to 0.40 (P = .03).
- Westwood, 2025: overall ADHD symptoms on probably blinded ratings, children only, 18 RCTs, 1126 participants, SMD 0.05 (−0.10 to 0.20).
- Westwood, 2025: inattention on probably blinded ratings, 19 RCTs, 1176 participants, SMD 0.04, −0.15 to 0.22.
- Westwood, 2025: hyperactivity and impulsivity on probably blinded ratings, 19 RCTs, 1176 participants, SMD 0.03, −0.09 to 0.16.
- Westwood, 2025: processing speed, k = 15; n = 909; SMD, 0.35; 95% CI, 0.01 to 0.69.
- Westwood, 2025: methylphenidate compared with neurofeedback on overall ADHD symptoms, k = 7; n = 417; SMD, −0.68; 95% CI, −0.92 to −0.44. What the terms mean:
- k The number of trials pooled in that analysis.
- n The number of people in those trials.
- SMD The size of the difference between groups on a common scale. Roughly, 0.2 is small, 0.5 is moderate and 0.8 is large. A minus sign shows the direction of the change, not its size.
- RCT A randomized controlled trial, where people are assigned to groups by chance.
- 95% CI The range where the true effect likely lies, so a range that includes zero means no clear effect was found.
- P How likely a result this size would be with no real effect, with values below .05 usually treated as unlikely to be down to chance.
- I2 How much the trials disagreed with each other, from 0% (not at all) to 100% (a great deal).
Source: Westwood S, Aggensteiner P, Kaiser A, et al.. Neurofeedback for Attention-Deficit/Hyperactivity Disorder. JAMA Psychiatry (2025). doi:10.1001/jamapsychiatry.2024.3702