Does creatine help the aging brain? What the trials actually show

Most people think creatine is now a brain supplement as well as a muscle one. The evidence is real — but it points somewhere much more specific than the marketing does, and once you line the trials up in order, the pattern is almost embarrassingly clear.

Creatine has had a very good few years. It went from “thing men in vests put in shaker bottles” to a supplement your GP might casually endorse, and somewhere in that rehabilitation it picked up a second job: cognitive enhancer. Sharper thinking. Better memory. Protection against the slow fog of an ageing brain. It’s on the tubs now, in that careful supplement-marketing tense — supports cognitive function — and it’s all over the podcasts.

I take creatine. I’ve written the honest case for creatine and healthy ageing, and I stand by it. But that post gave the brain evidence about four paragraphs and a shrug, and the shrug deserves a proper look, because the brain claim is not the same quality of evidence as the muscle claim. Not remotely. And the gap between them is where the interesting bit lives.

The claim, as everyone states it

The everyday version goes like this. Your brain is metabolically extravagant — about 2% of your body weight burning something like 20% of your energy. Creatine is an energy buffer. Therefore, more creatine, more available brain energy, better thinking. Especially as you age, when everything metabolic gets a bit creaky.

Each step in that chain is true-ish. The conclusion still doesn’t follow, and the reason is one most supplement copy quietly skips: an energy buffer only helps when the buffer is the thing that ran out.

Where it came from

The origin story is a single, genuinely lovely study.

In 2003, Caroline Rae and colleagues published a double-blind, placebo-controlled, cross-over trial in Proceedings of the Royal Society B. Forty-five young adults took 5 g of creatine daily for six weeks, and the creatine phase produced significant improvements on backward digit span (working memory) and Raven’s Advanced Progressive Matrices (fluid reasoning) — both timed, both processing-speed-dependent, both p < 0.0001. For a cheap white powder, that’s a headline. (According to PubMed: Rae et al., DOI.)

Here’s the detail that rarely survives the retelling: every participant was a vegetarian. Dietary creatine comes almost entirely from meat and fish, so vegetarians run measurably lower baseline stores. The foundational study of creatine-and-cognition was run in precisely the population with the most room to gain, and the result has been generalised to everyone ever since.

That’s not a scandal — it was a sensible design choice for detecting an effect. It’s just the first crack in the “creatine makes brains work better” sentence, and it turns out to be the same crack that runs through the entire literature.

What the strongest evidence actually shows

Let me steelman it properly, because there is a real signal here and the good studies deserve to be stated at full strength.

Sleep deprivation is where creatine looks best. Terry McMorris and colleagues ran two trials in which participants took 20 g/day for a week and were then kept awake. In the first (Psychopharmacology, 2006), after 24 hours without sleep the creatine group showed significantly less deterioration on random movement generation, choice reaction time, balance and mood. In the second (Physiology & Behavior, 2006), across 36 hours, creatine helped — but only on the complex central-executive task, and only at the 36-hour mark. (According to PubMed: McMorris et al., DOI; McMorris et al., DOI.)

Then, much more recently and much more elegantly: Gordji-Nejad and colleagues in Scientific Reports (2024) gave a single large dose — 0.35 g/kg, roughly 25 g for a 70 kg adult — to people during a sleep-deprived night, while scanning their brains. They measured actual changes in cerebral high-energy phosphates: phosphocreatine-to-inorganic-phosphate ratios shifted, ATP shifted, the drop in brain pH was blunted, and processing speed and cognitive performance improved. This is the study that impressed me most, because it doesn’t just report a test score — it shows the mechanism happening inside the skull. (According to PubMed: Gordji-Nejad et al., DOI.)

And in older adults, there’s at least one clearly positive trial. McMorris and colleagues again (Aging, Neuropsychology, and Cognition, 2007) gave elderly participants 20 g/day for a week and found significant improvements on nearly every task tested — number and spatial recall, random number generation, long-term memory. (According to PubMed: McMorris et al., DOI.)

So: real trials, real placebo groups, real effects, and one of them with brain imaging attached. Anyone telling you creatine does nothing for cognition hasn’t read these.

The pattern nobody names

Now put those studies in a row and ask a different question — not did it work, but who was it working in.

Sleep-deprived. Vegetarian. Elderly. In the broader literature, also: unwell.

Every clean positive result in creatine-and-cognition research comes from a brain that was already operating at an energy deficit. Not one of them shows creatine raising the ceiling of a rested, well-fed, omnivorous adult brain. That’s the synthesis the individual papers can’t state on their own, and it reframes the whole thing: creatine doesn’t appear to make brains better. It appears to make depleted brains less depleted.

The best single piece of evidence for that reading is hiding inside the most creatine-favourable meta-analysis in the field. Prokopidis and colleagues (Nutrition Reviews, 2023) pooled ten RCTs on memory in healthy people and found a modest overall benefit — standardised mean difference 0.29. Good news for creatine. But the subgroup split is the real finding: in participants aged 66-76, the effect was 0.88 — large. In participants aged 11-31, it was 0.03 — indistinguishable from nothing. Same supplement, same doses, same outcome measure; the entire pooled effect was carried by old brains. (According to PubMed: Prokopidis et al., DOI.)

Dose didn’t predict benefit. Duration didn’t predict benefit. Sex didn’t. Depletion did.

The honest verdict: a real effect, on a thin and contradictory base

Three things keep this from being a brain-health recommendation, and they’re the three you never hear on a podcast.

One: the best-designed long trial in older adults is null. Alves and colleagues (PLoS One, 2013) ran 24 weeks — not one week — in older women, randomised across four arms: creatine, placebo, and each with or without strength training. It’s the trial the field needed: proper duration, proper factorial design, a comprehensive battery covering memory, selective attention and inhibitory control. The result: no cognitive change from creatine at all. Strength training improved mood and strength; creatine added nothing cognitively, trained or untrained. (According to PubMed: Alves et al., DOI.)

Notice what separates that from the positive elderly study: 24 weeks versus one week, randomised parallel groups versus a sequential design where one group simply took placebo then creatine, and dozens of participants versus fifteen. The better-built study found less.

Two: the two meta-analyses on this literature disagree about who benefits. Xu and colleagues (Frontiers in Nutrition, 2024) pooled sixteen RCTs and found significant improvements in memory, attention time and processing speed — but no significant effect on overall cognitive function or executive function, and a subgroup pattern pointing the opposite way from Prokopidis: larger benefits in people aged 18-60 and in people with diseases. Two competent teams, largely overlapping trials, contradictory conclusions about the population that matters. (According to PubMed: Xu et al., DOI.)

When meta-analyses of the same evidence disagree on direction-by-subgroup, the correct read isn’t “pick the one you like.” It’s that the underlying trials are too few, too small and too heterogeneous to settle the question.

Three: the entire field is smaller than one mid-sized drug trial. Those sixteen RCTs in the Xu analysis contain 492 people in total. Not 492 per arm — 492 across the whole literature, spanning thirty years, healthy and diseased, teenagers to seventy-somethings. The muscle evidence for creatine runs to hundreds of trials and thousands of participants. The brain evidence is a rounding error next to it, and it is being sold with the same confidence.

There’s a mechanistic reason to expect exactly this, incidentally: the blood-brain barrier restricts creatine transport far more than muscle does. A standard 3-5 g daily dose saturates muscle within weeks and barely nudges brain stores — which is precisely why the study that needed an acute cerebral effect had to use about 25 grams in one go.

What to do instead

I’m not telling you to stop taking creatine. I still take mine. But here’s the framework I’d actually use, which is a good deal narrower than the tub:

Take creatine for muscle. That’s where the evidence is genuinely robust, and preserving muscle and strength into your seventies is one of the better-supported things you can do for how you age — the muscle case is the real case, and it’s worth comparing against your protein intake rather than treating either in isolation.

Expect a cognitive benefit only if you’re depleted. On the current evidence, the people plausibly likely to notice something are: vegetarians and vegans, people over about 65, people going through genuine sleep deprivation, and people who eat very little meat. If you’re a well-rested omnivore in your thirties expecting sharper thinking, the honest prediction from the data is: nothing detectable.

Don’t use it as a sleep substitute. The sleep-deprivation results are the strongest in the field and they’re also the most misread. Creatine partially blunted the damage of a night without sleep; it did not replace the sleep. Actual sleep duration and quality has a mortality evidence base creatine can only dream about.

So, plainly: creatine’s cognitive effect is real, small, and conditional on a brain that’s already short of energy — and the claim that it enhances a healthy adult mind rests on 492 people, two meta-analyses that disagree, and one trial in vegetarians from 2003.

Which, honestly, is a more interesting supplement than the one on the label. It’s just not the same one.