Does neurofeedback actually improve sport performance?
A 2025 meta-analysis is the most rigorous look we have so far at whether neurofeedback helps athletes. It pulled together 21 studies and 271 participants and asked one focused question: does EEG neurofeedback improve actual motor performance in sport? The answer was a moderate-to-large boost, with an overall effect size of Hedges' g around 0.78 (Pacheco et al., 2025) (open access, PMC).
For context, g around 0.78 sits close to a Cohen's d of 0.8, which is the boundary into a strong effect. In sport, where a couple of percentage points separate first place from fifth, that size of effect matters. If you compete in a precision sport like golf, archery, shooting, or darts, or you play a fast-execution sport like tennis or hockey, this is worth understanding in detail.
I run a weekly livestream where I work through new research and answer your questions live. This article is built from that session. I'm Andrew Hill. I'm a cognitive neuroscientist and brain-training coach, I founded Peak Brain Institute, and I've read more than 25,000 QEEG brain maps. Let me walk you through what the data say and how to use them without wasting your time.
What made this meta-analysis different?
We have seen neurofeedback sport research before. There was a 2018 review of 10 studies and a 2021 review of 13. Effects in older reviews tended to wash out, partly because they mixed in cognitive tests and questionnaires alongside the behavior that actually matters on the field. This 2025 analysis did three things differently.
First, it measured real-world motor performance only. Putting accuracy and consistency. Archery scoring at competitive distance. Rifle and pistol grouping and dispersion. Dart throwing. Free throws. Ice hockey shooting accuracy. Dance, dynamic balance, cycling time to exhaustion. These are mostly closed-skill precision tasks, self-paced, where you need fine motor control and the ability to stay out of your own way mentally.
Second, the authors scored study quality with the CRED-nf checklist, a consensus standard developed for neurofeedback research by Tomas Ros and colleagues (Ros et al., 2020). Think of it as quality control. Did the study report participant strategies? Document data processing and artifact handling? Justify the reinforcement schedule? Show regulation success in the feedback signal itself? And correlate brain change with behavioral outcome? That last point matters to me as a researcher. You always test behavior alongside the brain. The contrast between the two teaches you far more than either snapshot alone.
Third, study quality predicted the result. When you split the studies by CRED-nf score, the high-quality half showed a large effect of g = 1.07, more than double the g = 0.49 in the lower-quality half (Pacheco et al., 2025). Better-designed studies found bigger effects, not smaller ones. Most of those high-quality studies were published between 2022 and 2024, after the earlier reviews closed. The evidence base here is genuinely stronger than what we had a few years ago. You can read the journal page at Wiley and the PubMed record.
Who benefits most, novices or experts?
This is where the data get interesting. When the analysis split athletes by skill level, novices showed a large effect of g = 1.23. Experts dropped to about g = 0.6, which is a moderate, still-significant effect, though the overall novice-versus-expert split sat at a trend level (p = 0.06) (Pacheco et al., 2025).
That pattern makes mechanistic sense. If you are undertrained, there is more room to close the gap quickly. If you are already intermediate to advanced, you are closer to your ceiling, so the change is smaller. I see the same shape in the brain-training work. The worse things are at baseline, in attention, stress, anxiety, or sleep, the faster the first big shift tends to arrive. More room to improve means a bigger early boost.
A g = 0.6 for an expert is still meaningful. In elite sport, easing a one or two percent performance bottleneck is the difference between standing on the podium and not. Set your expectations by where you actually are. If you are a weaker performer with a big handicap, expect substantial gains. If you are already elite, expect smaller but consequential ones.
A real example: a hockey player after concussion
I worked with an NHL player who came in after an injury that pulled him out of play. We started with a QEEG brain map. For anyone who has not seen one, a QEEG measures how strong someone's resting brain patterns are compared with the average person their age. These are stable resource traits, not mood states. They show how the brain is allocating its activity across regions.
His map showed large amounts of theta and some delta and alpha across the back of the head, with relatively low beta. In plain terms, the slow brainwaves had taken over the visual system. That fits the picture of a concussion months out: visual tracking trouble, fogginess, fatigue, light sensitivity. His executive function testing told the same story. His attention was already exceptional, scoring 128, but his stamina was dropping, leaving him a little brittle and reactive under load. The visual processing speed sat lower than the auditory.
We did about 50 sessions of neurofeedback. Three months later his resting map looked substantially different. The theta, delta, and posterior alpha had come down. The beta had come up. On retest, the strong scores got stronger and the stamina problem eased, with his consistency rising sharply. That was around when he got signed and went back to work. With a high performer, the deficit may be small, but you can still see real performance change over a few months. The same logic applies if you are using neurofeedback to flatten an aging trajectory rather than recover from injury, watching reaction time, sustained attention, sleep, and alpha peak frequency get crisper over time.
How much training do you actually need?
This is the question I get most: how many sessions before something happens? The motor learning data give a useful answer. When neurofeedback is paired with the actual motor practice of the sport, the total training time to produce these effects is small, on the order of two or three hours, broken into 15 to 30 minute chunks. Several studies showed sport performance changes even after single sessions, likely through control over arousal and activation level.
For practical purposes, you can probably see a noticeable sport effect in two to three weeks, around six to ten sessions. That is roughly when we traditionally expect effects to emerge in sleep and executive function too.
I can show you why this works at the level of the EEG. In my doctoral research I trained people at C3 while recording from a 64-channel cap, and I measured the event-related spectral perturbation, the ERSP. This tracks how brain frequencies shift in the moment around each feedback beep. When someone trained C3 beta, you could see a clear blue stripe of beta change in the very first 10 minutes of the very first session. When someone trained SMR, the lower-frequency stripe showed up clearly by the end of the first half hour. The sham condition showed nothing. The brain reacts to real feedback fast, and those reactions get stronger and cleaner across sessions.
There is also a hemisphere story here. Training left-side C3 produced a 12 Hz echo over the opposite C4 electrode, in the homotopic frequency on the other side of the head. You get genuine cross-talk between hemispheres, with the strongest effect under the trained wire.
One caveat for measurement. You can get subjective and behavioral changes within a handful of sessions while the resting EEG has not budged yet. In my own pacing, I map at the start, and the resting brain often shows little change at sessions 12 to 15. By 18 to 20 sessions the resting changes are far more likely to appear, and by about 25 sessions they tend to bloom. That is why I map at intake and then about every 25 sessions, rather than chasing resting change too early.
How do you measure whether it's working?
Measure the sport, not just the EEG. The studies that scored highest on CRED-nf tied brain regulation to a real-world behavioral outcome, and that is the discipline to copy. Pick the actual mechanic of your sport and track it.
- Shot grouping or dispersion: measure it in centimeters or inches.
- Putting: measure percentage made or strokes.
- Consistency: log every run, every at-bat, every attempt and watch the variance.
- Reaction and impulsivity: a continuous performance test (go/no-go) reliably picks up neurofeedback change.
Other signals converge when you are doing the work well. Sleep trackers, HRV biofeedback devices, academic testing in younger people, and even medication sensitivity all shift. Stimulants can become noticeably too strong as neurofeedback starts to regulate attention. When everything points the same direction, you know the change is real rather than a story you are telling yourself.
For protocols, I am not going to prescribe one blind. Broadly this is an SMR approach tailored to the individual. You might train up alpha peak frequency for processing speed, or if you suffer from paralysis by analysis, you might inhibit theta on the frontal midline. Get a brain map first and target your training the way you would target a lipid panel or a bone density scan. Esports respond well here too, since impulsivity and sustained focus dominate that arena, and flow-state tuning becomes possible with alpha speed and alpha-theta work.
How strong is this evidence, really?
I want you to hold this honestly. The effect sizes are strong, and the high-quality studies are recent and well-designed. But there are limits worth naming.
There was real heterogeneity across studies, and even within studies across participants. Some athletes saw large gains, some modest, some none, and that variability does not show up in a single headline number. The sample sizes were small, averaging about 13 participants per study, so the effects are less stable than a thousand-person trial would give you. Replication and larger samples are needed before I would treat g = 1.07 as a fixed value.
Only about half the studies scored well on CRED-nf. Common gaps were not reporting whether participants got cognitive strategies, not documenting artifact correction, and not correlating regulation success with sport outcome. There is also a curious finding worth flagging. The sham-controlled studies showed weaker effects than the uncontrolled ones, which suggests attention and practice are inflating some of the apparent benefit. A 2024 attention meta-analysis raised the same sham-control caveat for cognitive outcomes, and a 2023 dose meta-analysis explored how training amount maps to motor performance.
So treat the 2025 result as the best evidence we have yet that neurofeedback moves sport performance, while keeping the placebo question open.
Is brain training for sport legal and ethical?
People ask whether training faster reaction times, lower impulsivity, deeper sleep, and more sustained attention counts as "neurodoping." Biofeedback and neurofeedback are legal in competition. The 2025 WADA Prohibited List does not ban training your own brain. You have spent years training your legs, your patience, and your read of the game. Training the brain directly is the same category of work.
If you compete, from chess to fencing to cycling to tennis, and you are performing below where you want to be, a brain map is a reasonable next measurement. Across my own coaching work I have trained race car drivers, soccer players, hockey players, NFL and football players, boxers, and competitive junior tennis players. Tennis in particular shows a clear effect on overreaction, hand-eye coordination, balance, and reaction time. I also send athletes to a batting cage to feel it, because hitting a baseball is about the most taxing combination of motor skill and brain timing there is. If you can hit reliably under that load, most fast-execution sports get easier.
Sensory sensitivities respond too. If you over-react to clothing texture or smell, that often traces to a cramped right temporoparietal junction, and sensory and social processing is a target neurofeedback tends to move well.
Where to start
If you want to test this on your own performance, start with a brain map so your training is targeted rather than generic, then pick the one sport metric you can measure precisely and log a clean pre-training baseline. Train two to three short sessions per week alongside your normal practice, and re-measure the sport metric, not just the EEG, at two to three weeks. The data say you can expect a noticeable change in that window, larger if you are earlier in your development and smaller but still useful if you are already elite. Peak Brain runs in-person mapping in several cities and fully remote neurofeedback across the US, UK, and Europe, so you can do this from anywhere.