Does Creatine Actually Help Your Brain? What 16 Trials Found

Somewhere in 2024, creatine stopped being a gym supplement and became a brain supplement, and one meta analysis of sixteen trials did most of the lifting. You’ve probably seen its numbers quoted: better memory, faster attention, quicker processing. What you probably haven’t seen is the correction, the commentary, and the regulatory opinion that followed.

Here is the short answer to whether creatine helps your brain. Sixteen randomized trials pooled in 2024 reported small average benefits on memory and on the speed of attention and processing tasks, and none on overall cognition or executive function. That pooled result has since been criticized for a statistical error that inflates precision, the European regulator reviewed the same evidence and found no established cause and effect relationship, and the largest single trial ever run found an effect too small to reach significance.

So creatine might help the brain a little, in some people, under some conditions. The evidence does not yet say more, and anyone telling you it does is ahead of the data.

I’ve been reading this literature for years. What follows is what the papers say, including the parts that don’t fit the marketing.

Why Creatine Might Matter For The Brain At All

The idea is not silly. The brain is about 2 percent of body weight and uses roughly 20 percent of the body’s resting energy, and creatine helps keep that supply steady. Cells hold part of their creatine as phosphocreatine, a charged form that can hand a phosphate to spent ATP and rebuild it in a fraction of a second, and the brain carries its own dedicated form of creatine kinase, the enzyme that runs the exchange.

People born unable to make or transport creatine into brain cells have profound cognitive problems, so the brain clearly needs it. Whether a brain with normal creatine gains anything from more is a different question.

A mechanism gets you a hypothesis, not a result. A 2025 perspective in The Journal of Nutrition by Igor Eckert makes the point with a cautionary tale: creatine once had an equally plausible mechanism as a neuroprotective agent, and one of the largest creatine trials ever run in a neurological disease, more than 1,700 patients followed for five years, was stopped early for futility. Plausible mechanisms fail in humans all the time.

Where The Number Sixteen Comes From

The sixteen trials come from the 2024 Frontiers in Nutrition meta analysis of 16 randomized trials by Chen Xu and colleagues. A meta analysis is a study that pools the results of earlier trials into one estimate. This one searched three databases for randomized controlled trials published between 1993 and June 2024, kept sixteen, and pooled 492 participants whose average ages ran from about 21 to 76.

Every trial used creatine monohydrate. Thirteen were in healthy adults and three were in patient groups.

The doses and durations were all over the map. Nine trials ran for less than four weeks and seven for longer. Daily doses ranged from roughly 2 grams to 20 grams, and durations ran from a single large dose during one sleepless night to 18 months of daily use.

Some trials tested people at rest, others under sleep loss, simulated altitude, exhausting exercise, or mental fatigue. Pooling all of that into one number is a choice, not a law of nature.

Here is what the pooled analysis reported. On memory, creatine came out ahead by a standardized mean difference of 0.31, a way of expressing an effect in units of spread so that different tests can be combined, with a 95 percent confidence interval from 0.18 to 0.44. By convention that is a small effect.

On the time taken to complete attention tasks, creatine came out 0.31 standard deviations faster, with an interval that only just excluded zero. On processing speed time the difference was 0.51 standard deviations, with a wide interval and 63 percent heterogeneity, a statistic meaning the trials disagreed with each other substantially.

It’s worth pausing on what 0.31 means. Pick two people at random, one on creatine and one on placebo, and the creatine taker scores higher on a memory test roughly 59 percent of the time, against 50 percent by chance. That is real across a population and nearly invisible in an individual.

Now the part that gets cut from the summaries. Overall cognitive function, executive function, attention scores as opposed to attention time, and processing speed scores all showed no significant effect. The authors graded their memory finding as moderate certainty and everything else as low or very low, and their own conclusion asked for larger trials.

The subgroups were stranger still. Benefits looked larger in the three patient groups than in healthy people, larger in adults aged 18 to 60 than in people over 60, and larger in women than in men on processing speed, the last drawn from just four studies. That people over 60 showed no attention benefit runs against the story most creatine content tells, and I’ll come back to it.

The Problem Nobody Puts In The Headline

Read the memory forest plot closely and something odd appears. The whole review contains 492 people, yet the memory analysis pools twenty four entries and lists 1,000 participants. Several trials reported many memory subtests from the same people, and each subtest was entered as if it were a separate study.

Two trials contributed at least seven memory outcomes each. Three others contributed four apiece.

This is a unit of analysis error, and it is not pedantic: treating correlated outcomes from the same participants as independent makes the pooled estimate look more precise than it is, shrinks the confidence interval, and pushes results across the significance line that shouldn’t cross it. A 2026 commentary in the same journal by Tom Citherlet of the University of Lausanne laid this out and asked for a reanalysis using multilevel models or averaged outcomes. I have not found a published reanalysis.

In 2022 a separate team published the 2023 Nutrition Reviews meta analysis on creatine and memory, eight trials and 225 people, reporting a memory benefit of 0.29 standardized units overall and a much larger effect, 0.88, in adults aged 66 to 76. That paper became the citation behind a thousand product pages.

Then Igor Eckert and Eric Pascher wrote to the journal pointing out the same double counting. The authors agreed the criticism was valid, reran the numbers, and the overall memory effect was no longer statistically significant. Only the older adult subgroup survived, and that subgroup rests on two small studies.

So the two most cited positive meta analyses share the same flaw: one has been corrected, and the other has a published commentary asking for the same. The Xu paper also carries a February 2025 corrigendum, but that fixes a translation error in the attention section; the double counting is the real story.

What The Regulators Concluded

In November 2024 the European Food Safety Authority published EFSA’s scientific opinion on creatine and cognitive function, in response to an application from the creatine manufacturer Alzchem Trostberg. The applicant put forward 21 human intervention studies and, when asked for more, submitted the Xu meta analysis. The panel’s verdict was that a cause and effect relationship between creatine supplementation and improved cognitive function has not been established.

Two details in that opinion deserve attention. First, the panel independently flagged the same problem the 2026 commentary would raise, noting that the meta analysis pooled related cognitive tests from the same studies, double counted participants, and inflated sample sizes, and it concluded that nothing could be drawn from that analysis. Second, three of the field’s classic positive studies, the two McMorris trials from 2006 and 2007 and the Rae vegetarian trial from 2003, had already been assessed by EFSA in 2011 for a memory claim and were judged to have limitations including missing randomization detail, inadequate adjustment for baseline differences, and multiple uncorrected comparisons. The United Kingdom’s Nutrition and Health Claims Committee assessed the same claim separately and reached the same conclusion.

This is not an authority that reflexively rejects creatine, either: in 2011 it authorized the claim that creatine increases physical performance in successive bursts of short term, high intensity exercise, at 3 grams a day. When the trials are there, the claim gets through. For cognition, the panel decided they were not.

For readers in the United States, where this site is based, EFSA’s opinion binds nothing. American supplements are regulated by the FDA under a framework in which the agency does not approve claims in advance, and a supplement may not claim to diagnose, treat, cure, or prevent any disease. Advertising claims are policed by the FTC, whose Health Products Compliance Guidance says health benefit claims need competent and reliable scientific evidence, which as a general rule means randomized controlled human trials.

A US company doesn’t have to convince EFSA. It does have to substantiate what it says, and the trials are the same on both sides of the Atlantic.

The Trials Worth Reading One By One

Here are the individual trials that carry the most weight, with population, dose, and duration attached. The trial that started the vegetarian story is Rae and colleagues, 2003: 45 young vegetarian adults, 5 grams a day for six weeks in a double blind crossover design, meaning each person took both creatine and placebo in random order, tested on backward digit span, a working memory test, and Raven’s Advanced Progressive Matrices, an abstract reasoning test. The effects were large and highly significant. It is a beautiful result with one problem: a much bigger attempt to reproduce it did not.

That attempt is the 2023 BMC Medicine trial of 123 adults by Julia Sandkühler and colleagues in Germany, the largest trial of creatine and cognition ever run. It was preregistered, meaning the analysis plan was published before the data came in, double blind, placebo controlled, and crossover, with 5 grams a day for six weeks in each arm, and about half the participants were vegetarians.

On backward digit span the creatine effect missed significance, with a p value of 0.064, and on the Raven matrices it was not close. On eight exploratory tests there was no sign of benefit.

Vegetarians did not benefit more than omnivores. Side effects were reported about four times as often on creatine as on placebo. The authors’ Bayesian analysis found weak evidence for a small effect and strong evidence against a large one, which is as fair a summary of this field as exists.

Rawson and colleagues, 2008, gave 22 young adults about 2.2 grams a day for six weeks and found nothing on memory or reaction time. That is the nearest thing to a trial of the low daily dose many people actually take.

The McMorris trials are the ones people mean when they say creatine helps under stress. In 2006, nineteen young adults took 20 grams a day for seven days and were then tested through a period of sleep deprivation with mild exercise; the creatine group was reported to do better on several memory and reaction tasks. In 2007 the same group gave 20 grams a day for a week to 32 adults averaging 76 years old and reported better forward number recall, spatial recall, and long term memory, but no difference on backward recall or random number generation.

That 20 gram dose is four times what most people take, and the 2007 study did not explicitly describe how participants were randomized. In 2024 Terry McMorris himself led a systematic review in Behavioural Brain Research whose title says the research fails to support the theoretical basis for an effect on cognition, concluding that effects in unstressed people are small to absent and that future trials must measure brain creatine directly. When the author of the field’s favorite positive studies says that, it’s worth listening.

Then there is Alves and colleagues, 2013: 56 older women aged 60 to 80, 5 grams a day for 24 weeks after a five day loading phase, with and without strength training, in a randomized double blind design. No effect on any cognitive measure. This is the longest creatine and cognition trial in older adults, and it is null.

The Sleep Deprivation Result Everyone Quotes

If one study powered the 2024 to 2026 creatine wave, it is the 2024 Scientific Reports sleep deprivation trial from Forschungszentrum Jülich. Fifteen healthy adults, average age 23, took a single dose of 0.35 grams of creatine per kilogram of body weight, roughly 24 grams for a 70 kilogram person, at the start of a night in which they were kept awake for 21 hours. Compared with placebo, and measured with phosphorus and proton magnetic resonance spectroscopy, the creatine night showed brain phosphocreatine and pH changes consistent with better energy status, and cognitive performance and processing speed held up better over the 7.5 hours after the dose.

It’s a careful study, and the imaging is the most direct evidence we have that a large acute dose reaches the brain within hours. But look at what it is: one night, fifteen people, a dose no supplement label recommends. It says almost nothing about a person taking 5 grams every morning and hoping to think more clearly at work.

The same group published a follow up in Nutrients in April 2026, this time with 29 healthy adults and a lower single dose of 0.2 grams per kilogram, about 14 grams for a 70 kilogram person. Creatine again reduced the deterioration on logical, numerical, and vigilance tasks during 21 hours awake, but the effect was smaller, up to about 12 percent, the short term memory benefit seen at the higher dose did not appear, and the authors wrote that the memory effect looks dose dependent. At ordinary daily doses there is no acute sleep deprivation trial at all.

One inconvenient detail. The 2023 Nutrition Reviews meta analysis split its trials by whether participants were stressed by sleep loss, hypoxia, or exhausting exercise. Memory benefits appeared in the unstressed trials, at 0.43 standardized units, and were essentially zero, at 0.03, in the stressed ones. That is the opposite of the popular narrative, and a reminder that the stress story rests on a handful of small trials that do not all agree.

Who The Findings Apply To, And Who They Don’t

Older adults are the group most often told creatine will help their memory, and the evidence here is thinner than the marketing. A 2026 systematic review of creatine and cognition in older adults found six eligible studies covering 1,542 people. Four were cross sectional surveys that estimated dietary creatine from food recall, a design that cannot establish cause. Only one was a randomized controlled trial, the Alves study above, and it was null.

The single study rated good quality was that null trial, and three of the six were rated poor. I would put it more bluntly than the review’s authors did: the older adult claim rests on one seven day study from 2007 with unclear randomization, and against it stands a 24 week randomized trial that found nothing.

Vegetarians and vegans are the second group, on the logic that they eat almost no creatine. They do have lower muscle creatine, but the 2023 BMC Medicine trial noted that vegetarians and omnivores have comparable total brain creatine, and its own vegetarian participants gained no more than the meat eaters. The brain, it turns out, makes its own creatine and guards its supply.

Patient groups showed the largest effects in the Xu subgroups, but that pool is three small trials, and one of them combined creatine with coenzyme Q10, so its result cannot be attributed to creatine alone. Women appear in the same analysis with a larger effect on processing speed time, drawn from four studies and rated low certainty, and the 2026 sleep deprivation paper reported that women benefited more on some tasks. Those are hypotheses to test, not findings to act on.

Healthy young adults under normal conditions have the clearest answer, and it is close to no. The 2023 meta analysis put the memory effect in its younger subgroup, labeled 11 to 31 years and oddly including a trial in children aged 10 to 12, at 0.03, indistinguishable from zero, and the largest trial ever run found at most a small effect on one working memory test.

What The Studies Used Versus What People Actually Take

Most people taking creatine for the brain take 3 to 5 grams a day, because that is the dose that saturates muscle. Almost none of the positive cognition results come from that dose. The McMorris trials used 20 grams a day. The sleep deprivation trials used 14 to 24 grams in a single hit.

The two trials that did use 5 grams a day, Rae in 2003 and Sandkühler in 2023, disagree, and the bigger one found an effect too small to confirm. The one trial at about 2 grams was null. The trials that would tell you whether a maintenance dose helps in an ordinary week have mostly not been run.

Dose matters for a physiological reason. Creatine crosses into the brain through a transporter, slowly and to a limited degree compared with muscle. A 1999 spectroscopy study found brain creatine rose by about 8.7 percent after 20 grams a day for four weeks, a fraction of the change seen in muscle, and the brain relies heavily on creatine it synthesizes itself.

A 2026 review in The Journal of Nutritional Physiology by Jedd Pratt and colleagues added that the accuracy and reproducibility of brain creatine measurement remain a serious problem, that most cognition trials never measured brain creatine at all, and that the optimal regimen for raising it is unknown. That applies to every creatine product equally, whatever the label implies.

One more thing worth knowing. Several researchers behind the positive meta analyses hold advisory roles with creatine manufacturers and disclose it; one coauthor of the 2023 memory paper sits on the scientific advisory board of the company that later applied to EFSA for the cognition claim. That doesn’t make the work wrong, but it does mean the null trials deserve the same attention as the positive ones.

That coauthor, Darren Candow, told NutraIngredients after the EFSA decision that he considered it justified given how small the research base is, and that it was too early to conclude creatine improves cognition overall. That is a fair summary from someone with every incentive to say otherwise.

Safety, Briefly

The safety record is the strongest part of the creatine literature. The International Society of Sports Nutrition’s 2017 position stand reviewed studies using up to 30 grams a day for up to five years and found no compelling evidence of harm in healthy people. Against that, the 2023 BMC Medicine trial did record about four times as many side effect reports on 5 grams a day as on placebo, and the sleep deprivation studies used single doses far above anything studied for daily use.

If you have kidney disease, take medication that affects the kidneys, are pregnant, or are thinking about giving creatine to a child, talk to a doctor before starting. Nothing in this article is medical advice.

What Would Change The Picture

The question isn’t settled in either direction, so here is what would settle it. A preregistered randomized trial of a few hundred adults, at a realistic daily dose, running for months, with brain creatine measured by spectroscopy before and after and a standardized cognitive battery, would do more than every trial to date combined. A proper reanalysis of the 2024 data with the double counting fixed would tell us whether the memory signal survives. A direct comparison of daily dosing against a single large dose during sleep loss would tell us whether the acute effect is the only real one.

Until then, here is how I’d read the sixteen trials. Creatine has a strong safety record at ordinary doses, it’s cheap, and it might do something small for memory or processing speed, most plausibly when the brain is under energy stress and at doses higher than people usually take.

It is not a nootropic in any sense that has survived a large trial, the two headline meta analyses have been corrected or challenged on the same statistical error, and the regulator that has looked hardest concluded the case is not made. Does creatine help your brain? Maybe, a little, sometimes. If someone sells you more certainty than that, they are selling you something other than the evidence.