Neurofeedback: What Research Shows

What neurofeedback is

Neurofeedback is a form of biofeedback that uses brain activity. A sensor, most often an electroencephalogram (EEG) on the scalp, records the signal. A screen or sound then shows the person how that signal is changing, and the person practices shifting it. In the United States, the Food and Drug Administration (FDA) describes a biofeedback device in general terms as an instrument that provides a visual or auditory signal so that a person can consciously adjust certain body measures. [1]

This page covers the research evidence. For consumer headbands that only record signals, see consumer EEG devices. For the wider research setting, see neurofeedback research context.

ADHD: the best-studied use

Label: mixed, with no clear benefit on blinded ratings.

• A 2025 meta-analysis in JAMA Psychiatry pooled 38 randomized trials (2,472 participants aged 5 to 40). On probably blinded reports of ADHD symptoms there was no significant improvement. The authors concluded that neurofeedback did not appear to meaningfully benefit people with ADHD, clinically or neuropsychologically, at the group level. [2]
• An earlier 2016 meta-analysis found benefits only on ratings by people closest to the treatment, who were the least blinded. Effects were not significant with probably blinded ratings or in trials with active or sham controls. Only 4 studies checked directly whether participants had learned to change their brain signal. [3]
• A 2013 review reached the same pattern: effects shrank to nonsignificant when the best probably blinded assessment was used. [4]
• A 2022 meta-analysis found improvement in sustained attention on laboratory tests (Hedges' g = 0.32). That effect was much smaller (g = 0.05) when only studies that blinded participants were included. [5]
• A sham-controlled trial of a scanner-based version (fMRI neurofeedback) in 88 boys with ADHD found no significant group difference on the main symptom scale. [6]
• In adults with ADHD, a 2025 network meta-analysis found that the effects of non-pharmacological strategies were inconsistent across different raters. [7]

Why blinding matters: when parents, teachers or participants know a treatment is being given, they may expect improvement and rate it that way. Sham-controlled and blinded designs help separate the training itself from expectation. See placebo and expectancy effects.

Other uses

• Tourette syndrome. A small double-blind, sham-controlled crossover trial of scanner-based neurofeedback in 21 adolescents found a greater tic reduction after real than sham feedback. [8] This is promising but limited: it was a small study, it used a research scanner rather than a consumer headset, and it does not show that home devices work.
• Sport and performance. An umbrella review of psychological skills training for sport (which included neurofeedback among many approaches) reported that most findings were weak or non-significant and certainty was generally very low. [9] Label: insufficient evidence.
• Sleep, anxiety, depression, focus, "peak performance". Proper evidence not found. The PubMed searches I ran for neurofeedback did not turn up a readable review or sham-controlled trial for these uses, so no claim is made.

What device rules do and do not mean

FDA's regulation for biofeedback devices places them in Class II (special controls) and exempts prescription, battery-powered units for relaxation and muscle retraining from premarket notification. [1] That describes the device category. It is not an FDA finding that neurofeedback treats ADHD, and a sold product's intended use matters. Accuracy of a device and benefit of training are separate questions.

How to judge a neurofeedback claim

1. Which condition and age group were studied?
2. Was there a sham or active comparison, and were raters blinded?
3. Did the study show participants learned to change the signal, and did real-world symptoms change?
4. Is the "brain training" claim a clinical outcome, or only a change in a brain signal? See clinical outcomes vs brain-signal changes.