Creatine and the Brain: What the Evidence Actually Supports
Creatine's cognitive effects are real but small, and they show up mainly under strain — sleep loss, age, or a vegetarian diet. An honest evidence grade and dosing guide.
For thirty years creatine was a gym supplement. A tub of white powder next to the whey, bought by people who wanted to move more weight. Then, over the last couple of years, it started turning up in conversations that had nothing to do with barbells — sleep, memory, the particular fog that settles over a Tuesday afternoon when you have been awake since four.
I got curious for unglamorous reasons. There is a small child in my house, and there have been stretches where five broken hours counted as a good night. On those days my thinking does not fail dramatically. It just becomes expensive. Holding four things in my head while tracing a bug costs more than it should, and I catch myself reading the same line three times.
So does creatine help with that? The honest answer is a qualified yes, and the qualification is doing most of the work.
What creatine is actually doing in there
Your brain is roughly two percent of your body weight and burns something like twenty percent of your resting energy. It has almost no capacity to store fuel. Neurons run on ATP that has to be regenerated continuously, second by second, and the demand is spiky — a burst of activity in one region needs energy faster than blood flow can deliver it.
The phosphocreatine system is the buffer that covers that gap. Creatine sits in the cell holding a phosphate group, ready to hand it to spent ADP and regenerate ATP in a single fast step, without waiting for the slower machinery of glycolysis or the mitochondria to catch up. Think of it as a capacitor rather than a battery: small capacity, very fast discharge, sitting right where the load is.
This is exactly the same mechanism that lets you push out two more reps. The muscle version is just louder.
The complication, and it is the central one for everything that follows, is the blood-brain barrier. Muscle takes up creatine readily. Brain does not. Uptake depends on a specific transporter, and that transport step is slow and easily saturated. Supplement for a week and muscle creatine can climb twenty percent; brain creatine in the same person over the same week moves by a few percent at best. Some studies find changes so small they sit at the edge of what MR spectroscopy can measure.
That single fact explains most of the confusion in this area. It is why brain protocols use larger doses than muscle protocols, why they run for longer, and why the effects — when they appear at all — are smaller than the gym results would lead you to expect.
What the evidence shows, claim by claim
The 2026 reviews, including a meta-analysis in Frontiers in Nutrition, land in roughly the same place: real effects, modest size, highly conditional on who is taking it and what state they are in.
Here is my read of where each commonly-made claim actually stands. The grades are mine, and they are deliberately conservative.
| Claim | Evidence grade | What that means in practice |
|---|---|---|
| Improves cognition when sleep-deprived | Moderate | The most consistent finding in the literature. Replicated across several small trials. |
| Improves memory in adults over 60 | Moderate | Meta-analyses show a small but reasonably consistent benefit, larger than in young adults. |
| Improves cognition in vegetarians & vegans | Moderate | Biggest effects show up in people who start with the lowest stores. Mechanistically coherent. |
| Improves memory in healthy, rested young adults | Weak | Effects are small and inconsistent. Several well-run trials find nothing at all. |
| Helps with mental fatigue during long tasks | Weak to moderate | Suggestive, but outcome measures vary so much across studies that pooling is hard to trust. |
| Improves mood or helps with depression | Weak | Interesting early signals, mostly as an add-on to medication. Not established. |
| Slows or improves Alzheimer's disease | Very weak | One small uncontrolled pilot. Generates a hypothesis. Proves nothing. |
| Protects against long-term cognitive decline | Insufficient | No trial has run long enough to answer this. Anyone claiming otherwise is guessing. |
Notice the pattern. Creatine looks best exactly where the brain's energy buffer is under strain or under-stocked, and worst where it is already comfortable. That is not a marketing story. It is what you would predict from the mechanism, which is a reason to take the modest findings more seriously rather than less.
Who is most likely to notice anything
Baseline is the whole game. If your stores are already near the ceiling, adding more does very little, which is why studies on well-fed young omnivores keep coming back flat.
Vegetarians and vegans. Dietary creatine comes almost entirely from meat and fish. An omnivore eats roughly one to two grams a day without trying. A vegetarian eats close to zero and runs on whatever the liver and kidneys synthesise. This group has the most headroom, and the cognitive trials in vegetarians show the clearest results in the whole literature.
Adults over sixty. Both tissue creatine and the efficiency of the whole energy-buffering system decline with age. Meta-analytic estimates put the memory benefit in older adults somewhere around a small effect — noticeable in aggregate across a study population, subtle in any one person.
Anyone genuinely short on sleep. Not tired in the ordinary sense. Actually deprived — new parents, shift workers, people on call. Sleep deprivation drops brain ATP availability, which is precisely the deficit a phosphocreatine buffer is built to cover.
People eating very little protein overall. Same logic as the vegetarian case, one step less extreme.
If you are a rested omnivore in your thirties who sleeps seven and a half hours and eats meat most days, the honest expectation is that you will notice nothing. That is a legitimate outcome, not a failure of the protocol.
Brain dosing is not muscle dosing
This is where most people go wrong, usually by taking the standard gym dose and expecting a cognitive result from it.
| Goal | Typical daily dose | Time to effect | Notes |
|---|---|---|---|
| Muscle saturation (standard) | 3–5 g | 3–4 weeks | Well established. Loading phases are optional and mostly speed things up. |
| Muscle, with loading | 20 g for 5–7 days, then 3–5 g | ~1 week | Same endpoint, reached sooner. More GI upset. |
| Brain, lower-dose trials | 5–10 g | 4–8 weeks | Mixed results. Some studies find no measurable change in brain creatine. |
| Brain, higher-dose trials | 10–20 g, or ~0.14 g/kg | 4–8 weeks | Where most of the positive cognitive findings come from. |
| Alzheimer's pilot | 20 g for 8 weeks | 8 weeks | Small, open-label, uncontrolled. Read it as a hypothesis, not a result. |
Two things follow from this table. First, the doses that produced cognitive findings are often three to five times the dose on the back of the tub. Second, nobody has established the long-term safety of twenty grams a day, because no trial has run long enough. Three to five grams daily has decades of safety data behind it. Twenty grams for eight weeks has eight weeks of data behind it.
That asymmetry deserves more weight than it usually gets in the enthusiastic version of this conversation.
What "modest" means when you are the one taking it
Researchers report effects in standard deviations. Most cognitive creatine findings land somewhere around a fifth to a third of one. It is worth translating that out of statistics, because the translation is sobering.
An effect that size means: if you lined up a hundred people on creatine against a hundred on placebo and tested their working memory, the creatine group's average would be slightly higher. It does not mean any individual in that group would feel different. Most would not. The signal is real and it is smaller than the noise of your own day-to-day variation.
The more useful framing is that creatine appears to raise a floor rather than a ceiling. On a well-slept day, when your buffer is full, there is nothing for it to do. On the fourth night of broken sleep, when the buffer is drained, it may return some fraction of what you lost. Nobody in the literature is describing anything resembling a new gear.
If you want the intervention that reliably fixes sleep-deprived cognition, it is sleep. Creatine is at best a partial hedge for the nights when sleep is not on offer.
Where the evidence is genuinely thin
The Alzheimer's pilot is worth naming precisely, because it gets cited far beyond what it can carry. Around twenty participants, high-dose creatine over eight weeks, open-label, no control group. Working memory improved. So did brain creatine on imaging.
With no control group, there is no way to separate the supplement from practice effects on repeated testing, from regression to the mean, or from the attention that comes with being enrolled in a trial at all. The researchers know this and say so. It is a well-designed pilot doing exactly the job a pilot is meant to do — establishing that a larger controlled trial is worth funding. It is not evidence that creatine treats Alzheimer's, and it should not be presented as such.
The broader literature has its own limits. Trials are small, typically twenty to fifty people. Outcome measures differ so much between studies that pooling them requires generous assumptions. Null results in supplement research have a long history of not being published. And the trials that do exist run for weeks, while the claims people make on the back of them run for decades.
If you want to try it anyway
None of this is medical advice, and I am an engineer rather than a clinician. But the practical picture is straightforward enough.
Creatine monohydrate is the form with essentially all of the research behind it. The exotic variants cost more and have not demonstrated an advantage. Three to five grams a day is the well-supported dose. If you are specifically after the cognitive effects and you have made peace with the thinner safety record, the trial protocols sit higher, and that is a decision to make with a doctor rather than with a forum thread.
Loading is unnecessary unless you are in a hurry. Timing does not matter. Take it with water and keep total fluid intake reasonable.
Two practical notes. Creatine supplementation raises serum creatinine, the marker used to estimate kidney function, without any change in actual kidney function. If you have bloodwork coming up, tell whoever ordered it that you are supplementing — otherwise you may get an alarming phone call about a number that means nothing. And if you have existing kidney disease, ask a doctor first. In healthy people the kidney concerns have not held up under study, but "healthy people" is doing real work in that sentence.
Give it at least four weeks before you judge anything, because brain uptake is slow. And judge it honestly. The temptation with any supplement is to notice the good days and forget the others.
Questions people actually ask
Will I notice anything? Probably not, if you sleep well and eat meat. Reasonably possibly, if you are vegetarian, over sixty, or genuinely sleep-deprived. Even then, expect subtle rather than obvious. Anyone describing a dramatic change is describing something creatine has not been shown to do.
Do I need to load? No. Loading reaches saturation faster and causes more stomach trouble. Three to five grams a day gets to the same place in three or four weeks.
Is it safe long term? At three to five grams a day, it is one of the better-studied supplements in existence, with decades of data in healthy adults. At the higher doses used in the cognitive trials, the long-term data does not exist yet. Those are two different questions and it is worth keeping them separate.
Will it make me gain weight? Some water retention inside muscle cells, commonly a kilogram or two early on. It is intracellular water, not fat, and it plateaus.
Does it work better with caffeine? There were old reports of interference, and they have largely not replicated. Most people take both without any apparent problem.
What strikes me most, reading through all of this, is how ordinary the mechanism turns out to be. No exotic pathway, no receptor being cleverly modulated. Just a cell running short of energy at the moment it needs energy most, and a small molecule that happens to be standing nearby holding a spare phosphate. Most of the honest answers in health turn out to look like that.