Somewhere in the last two years, creatine changed jobs.
For three decades it was a gym supplement — cheap, boring, unusually well proven. You took it, you lifted slightly more, nobody argued. Then the pitch shifted. Creatine became a brain supplement, sales spiked, and the same tub that used to promise a better deadlift started promising sharper thinking.
I went to check where that came from. What I found was not a marketing invention. There is a real body of research, published in real journals, and it does report benefits for memory. There is also a paper trail showing that a reader looked closely at the most important of those studies, found an arithmetic problem, and that the authors agreed with him.
This is what happened after that.
Educational, not medical advice. Creatine is well tolerated in healthy people, but anyone with kidney disease, on nephrotoxic medication, or pregnant should speak to a clinician first.
The paper everyone cites
In 2023, Nutrition Reviews published a meta-analysis by Prokopidis and colleagues pooling randomized placebo-controlled trials of creatine and memory in healthy people. Its conclusion was positive, and it became the citation. A great deal of what you have read about creatine and your brain traces back through it.
Meta-analyses carry weight for good reason. Individual trials are small and noisy; pooling them is supposed to reveal the signal underneath. The technique is only as sound as its bookkeeping, though, and bookkeeping is where this one had a problem.
January 2023: two researchers check the arithmetic
Igor Eckert and Eric Pascher wrote to the journal with a technical objection that turns out to be easy to understand.
Several of the pooled trials measured memory using multiple subtests on the same participants — one study contributing seven, another four, another four. Each subtest had been entered into the analysis as though it were a separate, independent result.
Do that, and the total number of observations exceeds the number of people who were actually randomized. The analysis behaves as if it has more data than it has. That artificially tightens the confidence intervals and inflates statistical power, and inflated power manufactures significance out of noise. The name for it is double-counting.
The reply
Here the story stops being ordinary.
Prokopidis and his co-authors published a response, and they did not defend the analysis. Their reply states that Eckert and Pascher “correctly identified a flawed statistical analysis,” acknowledges that the double-counting could have altered the outcome estimates and increased the risk of false positives, and reports a reanalysis using a single composite score per study.
Then they published what the corrected numbers showed.
The overall effect of creatine on memory was no longer significant. It survived in one place only: older adults. The sensitivity analyses in that same reply read as a row of nulls — an effect size of 0.22 at P = 0.15, 0.29 at P = 0.14, 0.03 under stressed conditions.
That is science working. A claim went out, a reader checked it, the authors conceded, the finding shrank to a subgroup. Nobody behaved badly. The correction is public and easy to find.
It is also almost never mentioned.
November 2023: someone runs the experiment properly
While that exchange was happening, a separate team was addressing the question from the other direction.
The most impressive single result in this field came from Rae and colleagues: 45 young vegetarian adults, placebo-controlled, double-blind, showing a large effect on both working memory and abstract reasoning at p < 0.001. It is the study people point to when they want one study.
So a group set out to replicate it — preregistered, cross-over, double-blind, placebo-controlled, 123 participants, the largest trial on creatine and cognition ever conducted, 5 grams daily for six weeks per arm. Roughly half vegetarians and half omnivores, deliberately, to test whether the vegetarian responder effect was real.
Their results, published in BMC Medicine:
- Backward Digit Span: p = 0.064. Close. Not significant.
- Raven’s Advanced Progressive Matrices: p = 0.327. Null.
- All eight additional exploratory cognitive tests: no indication of improvement.
- Vegetarians did not benefit more than omnivores.
The authors reported, honestly, that Bayesian analysis supported a small beneficial effect. But the headline finding did not replicate, and the subgroup that was supposed to explain everything didn’t behave as predicted.
The vegetarian idea deserves a moment, because it’s the standard fallback whenever creatine’s cognitive evidence gets questioned. It has genuine support: Rae’s trial was in vegetarians; a 2011 study of 128 young women found better memory in vegetarians and not meat-eaters, while finding no effect at all on verbal fluency or vigilance; a 2018 systematic review saw the same asymmetry. But the largest and best-designed test contradicts it, and there’s a mechanical problem underneath. Vegetarians have lower creatine in blood and muscle — and comparable total creatine in the brain. The deficit the hypothesis needs may not exist in the organ that matters.
2024: it happens again
The following year, Frontiers in Nutrition published a new systematic review — 16 trials, 492 participants — reporting benefits for memory, attention time and processing speed. It also found no significant effect on overall cognitive function, none on executive function, and rated its own evidence quality low-to-moderate. Those three facts rarely make it into the summaries.
In February 2026, a commentary in the same journal documented that the new analysis contained the identical error.
It names the studies: two by Alves, each contributing at least seven memory subtests; McMorris 2006, four; McMorris 2007b, four; Pires 2020, four — all treated as independent observations. The line to remember is the commentary’s own: the number of observations in the pooled analysis exceeds the number of unique randomized participants.
Same flaw. Identified, conceded, corrected — and then repeated, three years later, in a different journal.
For completeness: that 2024 paper also needed a corrigendum in early 2025 to fix an error in its attention section, attributed to a Chinese-to-English translation problem.
Why the effect keeps being so hard to find
None of this means creatine does nothing to a brain. It means the effect is small enough that how you count determines whether you see it. And there is a physical reason to expect exactly that.
Creatine cannot drift into the brain. It requires a transporter protein called CT1, and CT1 is sparse in the capillaries of the blood-brain barrier. Your brain manufactures its own creatine and shows little interest in imports — there is evidence that when you supply it from outside, internal production decreases in compensation, holding levels roughly where they were.
The numbers are stark. Supplementation raises muscle creatine by about 20%. Brain creatine rises by roughly half that.
And even that requires doses nobody takes. In 1999 Dechent and colleagues put six volunteers through four weeks of 20 grams a day and measured brain tissue directly with magnetic resonance spectroscopy. Total brain creatine rose 8.7%.
Twenty grams daily for a month, under nine percent.
A later study measured frontal-lobe phosphocreatine across 2, 4 and 10 grams daily and found increases of 4.6%, 4.1% and 9.1% — doubling from 2 to 4 grams achieved nothing at all. There is also evidence that sustained high dosing prompts the brain to reduce available transporters, which makes very large long-term doses partly self-defeating.
Now assemble the picture. Trials use the muscle-appropriate 5-gram dose. That dose barely moves brain creatine. Most studies never measure brain creatine, so nobody checks whether the compound arrived — a January 2026 review calls this absence widespread and says it has significantly impeded understanding of the whole question. The resulting small, inconsistent effects then get pooled using statistics that overstate their precision.
If you read our piece on krill oil and fish oil, this is the same architecture: a supplement that raises blood levels beautifully and brain levels barely, because the barrier has one narrow gate. Raising your muscle creatine and raising your brain creatine are different achievements.
Where it does work
Two contexts survive scrutiny, and both fit the mechanism rather than fighting it.
Acute metabolic stress. A 2024 study gave a single enormous dose — 0.35 g/kg, around 25 grams for a 70-kilogram adult — during 21 hours of sleep deprivation, with MR spectroscopy running throughout. It found real changes in cerebral high-energy phosphates, prevented a drop in brain pH, and improved cognitive performance and processing speed. Note the design: the researchers used a massive single dose specifically to force creatine past the transporter bottleneck. When brain energy metabolism is genuinely compromised, a buffer helps. When it isn’t, there is nothing to rescue — which is why six weeks of moderate or high-dose creatine in rested young adults produced no cognitive improvement and no change in prefrontal activation, and why a systematic review concluded that performance stayed unchanged in young individuals.
People with a deficit. The corrected meta-analysis kept its effect in older adults. The 2024 analysis found larger benefits in people with disease than in healthy participants. The only human trial in Alzheimer’s disease — a pilot in 20 patients on 20 grams a day for eight weeks — reported cognitive improvements alongside an 11% rise in brain creatine.
We keep arriving here. With magnesium, with blueberries, now with creatine: the benefit concentrates in people with something to correct, and disappears into the average once everyone else is counted. A healthy, rested adult under sixty is the population these trials most reliably find nothing in.
What creatine is not guilty of
Two things it gets accused of are genuinely myths, and the way each became folklore is worth seeing.
Hair loss traces to a single study of college rugby players taking 25 g/day for a week then 5 g/day for two more, whose serum DHT rose 56% and stayed 40% above baseline. DHT drives male pattern baldness, so the inference wrote itself. It has never been replicated; twelve subsequent studies of creatine and testosterone found no significant hormonal changes. In 2025 a randomized, double-blind, placebo-controlled trial finally measured the hair itself — density, follicular unit count, cumulative thickness, plus DHT and its ratio to testosterone — across twelve weeks at 5 g/day. Nothing moved.
One caveat, stated because we would state it if the result had gone the other way: that trial appeared in the journal of the International Society of Sports Nutrition, a trade body with close industry ties, written by authors affiliated with it. It converges with a decade of null hormonal data and I believe it. The conflict still belongs in the sentence.
Kidney damage has a mechanism behind the confusion. Creatine raises serum creatinine — the exact marker used to estimate kidney function. Higher creatinine reads as lower eGFR, which reads as damage. It is the supplement showing up in the assay. Trials running to five years find no renal impairment in healthy people. Real cautions are narrower: existing kidney disease, nephrotoxic medications, pregnancy and breastfeeding where research is simply absent. If you supplement, tell your doctor before a blood panel.
And one number that belongs here rather than in a footnote, because no enthusiastic write-up includes it. In that 123-person trial, participants reported side effects significantly more often on creatine than on placebo: p = 0.002, relative risk 4.25. Mostly gastrointestinal, mostly manageable, clustered at higher intakes. Not dangerous. But “no side effects” is not what the data says.
What I’d actually do
Take creatine monohydrate, three to five grams daily, for the reason it has thirty years of evidence behind it: strength, lean mass, training capacity. Every trial in the cognitive literature used monohydrate; it’s also the cheapest form, and comparative pharmacokinetics show the premium versions don’t change absorption. Loading — 20 grams a day for five to seven days — reaches saturation faster and is otherwise pointless, and it’s when the stomach complaints happen. Timing is irrelevant; this fills a reservoir over weeks.
If your goal is specifically the brain, the honest answer is that the doses which measurably raised brain creatine were 10 to 20 grams a day, that even 20 grams for four weeks moved it under 9%, and that nobody has characterised what sustained dosing at that level does to the transporters over years. That is an open question, not a protocol.
And be clear with yourself about the second effect. If creatine helps you train harder, and training drives the neurotrophic changes — the proposed pathway for creatine and BDNF runs through exercise, via muscle signalling, not through creatine acting on neurons — then the brain benefit you’re buying is the workout’s. Which is fine. It is just a different purchase than the one on the label.
The part that outlasts the supplement
I don’t think creatine’s cognitive story is over. My guess is it resolves in older adults, at doses higher than the standard scoop, in people whose brain energy metabolism is already under strain — which is where every surviving signal currently points, and where the next serious trials are heading.
What I’m more confident about is the smaller lesson.
Two of the most-cited meta-analyses in this field inflated their own certainty the same way, the second one three years after the error had been publicly identified and accepted by the researchers who made it. Nobody lied. Nobody hid anything. The correction is a public document with a DOI. It simply never travelled, because corrections don’t, while the original finding is still being cited in the form it had before anyone checked.
So when you next read that studies show something improves memory, the question worth asking isn’t whether the studies exist. They usually do. Ask whether anyone went back and checked the arithmetic — and whether, when someone finally ran the experiment properly, the finding was still there.
About this article
Written by Drew Anton. Covers nootropics, stimulants, sleep and focus protocols, and wearables. Not a physician or research scientist — reads the primary literature closely and refuses to round up.
Medical review: None. NeuriFuel does not currently have a licensed clinician on the editorial team, and this article has not been medically reviewed. We state this rather than implying an authority we do not have. See our About page for our full methodology.
Sources: Built from the primary meta-analyses together with their published corrections, commentaries and author replies; the largest randomized replication trial; magnetic resonance spectroscopy dosing studies; and the 2025 hair-follicle trial. Null results are reported alongside positive ones, and industry affiliation is flagged next to the study it affects — including where that study’s conclusion is one we agree with. Where a value could not be verified it is marked “data not specified.”
Corrections: Found an error? Write to hello@neurifuel.com with a source and we will fix it and log the correction.
Last updated: 29 July 2026
References
- Prokopidis K, Giannos P, Triantafyllidis KK, Kechagias KS, Forbes SC, Candow DG (2023). Effects of creatine supplementation on memory in healthy individuals: a systematic review and meta-analysis of randomized controlled trials. Nutr Rev 81(4):416–427
- Eckert I, Pascher E (2023). Letter to the Editor: Double-counting due to inadequate statistics leads to false-positive findings. Nutr Rev 81(11):1495–1496. DOI 10.1093/nutrit/nuac108; PMID 36644917
- Prokopidis K, et al. (2023). Author’s reply. Nutr Rev 81(11):1497–1500. DOI 10.1093/nutrit/nuac111 — concedes the flaw; effect non-significant after correction except in older adults
- The effects of creatine supplementation on cognitive performance — a randomised controlled study (2023). BMC Medicine. PMC10647179 — N=123, preregistered cross-over RCT, failed replication, side effects RR 4.25 (p=0.002)
- Xu C, Bi S, Zhang W, Luo L (2024). The effects of creatine supplementation on cognitive function in adults. Front Nutr 11:1424972. DOI 10.3389/fnut.2024.1424972; PMC11275561
- Xu C, et al. (2025). Corrigendum. Front Nutr. DOI 10.3389/fnut.2025.1570800; PMC11873458
- Commentary: Double-counting in “The effects of creatine supplementation on cognitive function in adults” (2026). Front Nutr. DOI 10.3389/fnut.2026.1716285
- Creatine Supplementation for Cognition: A Critical Perspective on Promise, Proof, and Public Perception (2025). J Nutr. DOI 10.1016/j.tjnut.2025.07.016
- Creatine supplementation and brain health: methodological challenges, current evidence, and translational perspectives (2026). J Nutritional Physiology. DOI 10.1016/j.jnphys.2026.100017
- Dechent P, et al. (1999). Increase of total creatine in human brain after oral supplementation of creatine-monohydrate. Am J Physiol 277(3):R698
- Kondo DG, et al. Frontal lobe phosphocreatine at 2, 4 and 10 g/day, ³¹P-MRS — retrieved via secondary review
- Gordji-Nejad A, et al. (2024). Single dose creatine improves cognitive performance and induces changes in cerebral high energy phosphates during sleep deprivation. Sci Rep 14(1):4937; PMC10902318
- Avgerinos KI, et al. (2018). Effects of creatine supplementation on cognitive function of healthy individuals: a systematic review of RCTs. Exp Gerontol 108:166–173. PMID 29704637
- Benton D, Donohoe R (2011). The influence of creatine supplementation on the cognitive functioning of vegetarians and omnivores. Br J Nutr — N=128, 20 g/day × 5 d
- Lak M, Forbes SC, Ashtary-Larky D, et al. (2025). Does creatine cause hair loss? A 12-week randomized controlled trial. J Int Soc Sports Nutr 22(sup1):2495229. PMID 40265319 — published in the ISSN’s own journal by ISSN-affiliated authors
- de Souza E Silva A, et al. (2019). Effects of creatine supplementation on renal function: a systematic review and meta-analysis. J Ren Nutr 29(6):480–489. PMID 31375416
- Jäger R, et al. (2007). Comparison of new forms of creatine in raising plasma creatine levels. J Int Soc Sports Nutr 4:17
- Creatine supplementation and muscle-brain axis: a new possible mechanism? Front Nutr. PMC12325066 — the exercise-mediated BDNF pathway
- Smith et al. (2025) — pilot trial, 20 Alzheimer’s patients, 20 g/day × 8 weeks, brain creatine +11%