Creatine for the brain: what the meta-analysis found and how to test it on yourself
Updated: 2026-09-22
"Creatine for brain" is one of the fastest-growing supplement searches of 2026 (+118%), with "creatine for women" close behind (+22%). Marketing promises mental clarity and protection against sleep loss. There is one RCT meta-analysis in healthy people worth reading before buying — it does not support an effect for most readers of this text.
What the meta-analysis found
Prokopidis et al. (2023, Nutrition Reviews) pooled creatine RCTs in healthy people; 8 studies went into the memory meta-analysis. Overall, creatine improved memory versus placebo: SMD 0.29 (95% CI 0.04–0.53, I²=66% — substantial between-study heterogeneity). Dose (2.2–20 g/day) and duration (5 days to 24 weeks) did not change the result.
The key detail: the effect is carried entirely by older adults. In the 66–76-year-old subgroup, SMD was 0.88 (a large effect); in the 11–31-year-old subgroup, SMD was 0.03 (95% CI −0.14 to 0.20 — effectively zero, the interval spans both negative and positive values).
So is it useless if you're young?
For memory at rest — based on this meta-analysis, yes. For a healthy 20–35-year-old with normal sleep, an average group effect of zero on cognitive tests is the most honest reading of this data.
What the meta-analysis does NOT test: creatine under sleep deprivation, stress, or high mental load. The hypothesis that the effect is larger there is physiologically plausible (brain phosphocreatine gets depleted precisely under such load), but it is not established by the cited data. We call it unproven, not implied.
Dose and how to take it
The standard maintenance dose is 3–5 g/day of creatine monohydrate; a loading phase is not required (it only speeds muscle saturation by 1–2 weeks — it makes little difference for the brain timeline). The effect is cumulative: a reasonable self-test horizon is at least 2–3 weeks of daily intake.
How to test the effect on yourself
Health OS has a protocol called "Creatine 5g for Cognitive Function" (`focus.creatine_cognitive.v1`): 7 days of baseline without creatine, then 14 days of 5 g/day in the morning, with an evening rating of cognitive fatigue and focus. It compares phase means (baseline vs. intervention); success is a cognitive-fatigue drop of at least 1 point, with a minimum of 10 nights of data for confidence.
Pair the subjective rating with objective tests — PVT (mean reaction time and lapse count) and n-back (d-prime). Look separately at short-sleep nights (<7h by Oura) — if there is any effect at all in a healthy person, physiologically it should show up there first.
A trap the marketing doesn't mention
PVT and n-back both carry a learning effect — scores improve with repeated testing alone, with no supplement involved. Over a 21-day protocol that stacks on top of any real drug effect. Watch the difference in the rate of improvement between baseline and intervention, not "it got better" — and don't over-read one good evening.
FAQ
Is the effect different for women?
The meta-analysis tested sex as a covariate: it did not change the result — nor did dose, duration or region. Age predicts the effect, not sex. These data show no separate 'female' effect in either direction.
Does creatine cause hair loss?
Not covered here — we did not find data confirming or ruling this out within the sources cited in this guide. We assert neither.
Where do I start if I'm specifically interested in sleep deprivation?
The meta-analysis did not test this scenario. The most honest way to find out for yourself is to run the `focus.creatine_cognitive.v1` protocol and separately compare nights with <7h of sleep against ordinary nights, rather than relying on the average result for healthy young adults.