I take creatine and have for years, originally for the same reason most of my patients do: muscle, strength, body composition. Five grams a day, the same dose every trainer on the internet recommends, the dose that’s been studied more than almost any other supplement in history.

The question I’ve been sitting with is whether that dose is doing anything for my brain, and having now looked at the evidence properly, the answer is: probably not much.

Creatine has entered the longevity conversation in a big way over the past two years. Health podcasters are calling it a nootropic. Biohackers are stacking it with their morning protocol. A pilot trial in Alzheimer’s disease made headlines in 2025. If you follow health media at all, you’d be forgiven for thinking the science is settled. The science is genuinely interesting, but it’s a long way from settled, and the gap between what we actually know and what’s being claimed is worth walking through carefully.

Your brain accounts for roughly 2% of your body weight but burns about 20% of your resting energy. Every neuron runs on ATP (adenosine triphosphate), and each cell only holds a few seconds’ worth at any given time. When demand spikes, the cell needs a rapid ATP top-up system. That system is phosphocreatine.

Think of it as a “UPS” (an uninterruptible power supply) for your neurons. The mains power is your mitochondria grinding out ATP from glucose and oxygen. The UPS is phosphocreatine sitting there, fully charged, ready to hand over its stored energy the moment demand exceeds supply. For a tissue that can’t afford even brief power interruptions, this buffer matters enormously.

The enzyme that manages this exchange is creatine kinase. In the brain, the specific isoform is CKB (brain creatine kinase), which sits almost exclusively in astrocytes, the support cells that feed and protect neurons (1). When CKB is working well, the phosphocreatine shuttle keeps energy flowing. When it isn’t, neurons start running brown-outs.

The brain both synthesises its own creatine and imports it from the bloodstream. The import route crosses the blood-brain barrier via a transporter called SLC6A8, and that transporter has limited capacity (2). Muscle tissue saturates with creatine supplementation in a few weeks. The brain takes much longer, responds less dramatically, and may need considerably higher doses to show measurable changes on MRI spectroscopy. This is the central fact that explains a lot of the confusion in the literature: the brain is simply harder to load than muscle.

The two findings from the past two years that changed how I think about creatine for my patients are a sleep deprivation study and an Alzheimer’s pilot, and they pull in different directions depending on how carefully you read them.

Gordji-Nejad and colleagues at Forschungszentrum Jülich in Germany published a clever trial in 2024 (3). They took 15 healthy adults, kept them awake for 21 hours, and gave them either a single high dose of creatine (0.35g per kilogram of body weight, roughly 25 to 30 grams for a 75-kilogram person) or placebo in a double-blind crossover design. The creatine group preserved their brain phosphocreatine and ATP levels. Their brain creatine increased on MRI. Their working memory and processing speed held up better than placebo. All of this within hours of a single dose.

The proposed mechanism is pharmacologically elegant: flood the bloodstream with creatine while the brain is under metabolic stress from sleep deprivation, and you temporarily open a window for faster brain uptake via SLC6A8. High supply meets high demand. A follow-up study from the same group in 2025 showed that creatine also corrects the hemispheric asymmetry in high-energy phosphate consumption that develops during sleep deprivation, essentially rebalancing left-right brain energy distribution (4). A separate replication using a lower dose (0.2g/kg in 29 participants) confirmed the cognitive attenuation effect, though with a smaller signal, suggesting dose-dependency.

This maps onto a very specific patient stereotype, which happens to represent most of my practice: 45-to-50-year-olds running big businesses, sleeping six hours on a good night, and asking their brain to perform complex executive tasks all day. They aren’t vegetarian university students (the population in most early creatine trials), nor are they 75. They are chronically sleep-compromised, cognitively loaded professionals who have exactly the metabolic stress profile where this intervention has the strongest mechanistic rationale.

The CABA trial (Creatine to Augment Bioenergetics in Alzheimer’s) was the first human study of creatine supplementation in Alzheimer’s disease (5). Smith, Taylor, and colleagues at the University of Kansas gave 20 patients with probable Alzheimer’s dementia 20 grams of creatine per day for 8 weeks. No placebo group, which I’ll come back to. The results revealed that 19 of 20 patients achieved excellent compliance. Brain creatine increased by 11% on MRI spectroscopy (p < 0.001). Cognitive scores improved on global cognition, fluid cognition, list sorting, oral reading, and attention tasks.

The 11% brain creatine increase is the finding I pay most attention to, because it’s a biological signal measured on imaging, independent of cognitive test scores, which in a 20-person uncontrolled trial could reflect practice effects. It tells us that at 20g/day, creatine actually gets into the brain in measurable quantities, and that matters because many earlier trials used 5g/day and likely never moved the needle on brain stores at all.

The foundational cognitive trial is Rae and colleagues from the University of Sydney in 2003: 45 young vegetarian adults, 5g/day for six weeks, large effect sizes on working memory and reasoning (6). It was a good study, but it was a study in vegetarians, who have lower baseline brain creatine because they get almost none from their diet. When Sandkühler and colleagues attempted to replicate this in 2023 with 123 participants including omnivores, working memory only bordered significance (p = 0.064) and reasoning showed no benefit at all (7). The largest effect in the original trial was likely filling a dietary deficit, not enhancing a system that was already adequately supplied, and that distinction matters more than most commentators have acknowledged.

The meta-analyses tell a similar story when you read past the abstracts. Avgerinos and colleagues in 2018 found modest improvements in short-term memory and reasoning across six RCTs, with stronger effects in vegetarians and under stress (8). Prokopidis and colleagues in 2023 pooled eight memory trials and found the effect was driven almost entirely by older adults aged 66 to 76, with a standardised mean difference of 0.88 versus 0.03 in younger adults (9). Essentially zero in the young. Xu and colleagues in 2024 reported improvements in memory and processing speed across 16 RCTs, though they found no significant effect on overall cognitive function or executive function, and the GRADE certainty of evidence was low for most outcomes (10). A published critique of the Prokopidis meta-analysis flagged that pooling non-independent cognitive tests from the same participants inflated the apparent sample sizes, a concern the European Food Safety Authority’s 2024 opinion confirmed. The overall effect size may be overstated.

The neurodegenerative disease trials are the most rigorous in the field and the most disappointing. The CREST-E trial in Huntington’s disease randomised 553 patients to up to 40g/day or placebo and was stopped early for futility, with the creatine group doing no better and possibly trending worse (11). The NET-PD LS-1 trial in Parkinson’s disease is the largest creatine trial ever conducted in neurology: 1,741 patients, 10g/day versus placebo, at least 5 years of follow-up, published in JAMA, and showing no slowing of clinical progression whatsoever (12). Pilot trials in ALS have also been negative. Across every large, rigorous neurodegenerative trial, creatine has failed to slow established disease. The CABA investigators argue that Alzheimer’s may differ because the energy deficit appears earlier in the disease cascade and their dose was higher, a reasonable hypothesis worth testing, but not evidence.

I should mention Sergej Ostojic, because two of the more optimistic recent papers come from his group: one linking dietary creatine to younger biological age on epigenetic ageing clocks (13), the other associating at least 1g/day dietary creatine with 15% lower all-cause mortality over 20 years of NHANES follow-up (14). Both sound encouraging until you look closely. The ageing clock correlations were tiny (r = -0.041 and -0.047), the kind of signal that reaches statistical significance because of sample size rather than biological magnitude. The mortality association weakened when demographic and lifestyle factors were included in the model, which matters because dietary creatine intake correlates directly with meat consumption, which correlates with socioeconomic status, diet quality, and a dozen other healthy behaviours. You’re largely measuring the lifestyle, not the creatine.

Ostojic’s conflict of interest disclosure is worth reading in full. He serves on the Scientific Advisory Board of AlzChem LLC, a creatine manufacturer. He co-owns a European patent on liquid creatine supplements. He has filed a patent application on creatine for telomere lengthening. He has received research support from multiple creatine-related companies. He is co-founder of KRE-ALL, a creatine food products company. I’m not dismissing his work; he’s published extensively and the data is the data. When a researcher with creatine patents publishes observational data suggesting creatine extends life, you weigh that differently than the same finding from an independent group. Any of my medical colleagues reading this would do the same. (In the interests of transparency, I am co-founder of Kurk - a bioscience company that specialises in curcumin-based supplements).

Candow and colleagues (some of the most credible researchers in this space, including Gordji-Nejad himself) published a review in 2026 titled “Creatine Supplementation and the Brain: Have We Put the Cart Before the Horse?” (15). They questioned whether the field has overclaimed the cognitive benefits, noting that many positive trials used doses inadequate to move brain creatine and that effect sizes are smaller than often portrayed. When the leading researchers in a field publish a paper questioning their own field’s enthusiasm, I pay attention!

So where does that leave me from a clinical standpoint? If I had to pick the person in my practice for whom creatine supplementation aimed at the brain is a no-brainer (sorry), it’s an older adult, 66 and above. The most consistent signal across every meta-analysis sits in that age group. Brain creatine synthesis declines modestly with age. The Prokopidis subgroup analysis showed a meaningful memory benefit (SMD = 0.88, p = 0.009) in this range (9). Brain energy economics narrow as we age, and supplementation may partially offset that decline. That said, the cost-benefit of creatine seems truly favourable at any age.

Vegetarians and vegans are the other group where the biology is straightforward. Dietary creatine comes almost exclusively from meat and fish. If you eat neither, you have measurably lower muscle and likely lower brain creatine stores. The Rae trial’s large effect sizes make biological sense in this population: you’re filling a genuine deficit, not trying to push an already-adequate system above baseline (6).

Then there’s the group I think most commentators miss entirely: the sleep-deprived, cognitively loaded professional. You don’t need to be vegetarian or 75 to benefit. You need to be running a brain under metabolic stress. Chronic sleep restriction, sustained high-level cognitive demand, and age-related decline in brain energy metabolism create exactly the conditions where the Gordji-Nejad mechanism applies (3). The gap between energy supply and energy demand is the therapeutic target.

For well-slept, meat-eating, 35-year-old gym-goers (the demographic that currently buys the most creatine), the evidence for cognitive benefit is weakest. The Prokopidis meta-analysis found essentially zero effect in younger adults (SMD = 0.03, p = 0.72). If you’re young, well-fed, and sleeping properly, your brain creatine system is probably working fine without help.

‍

Dosing

Five grams per day saturates muscle creatine stores within a few weeks. It’s the dose in most supplement recommendations, most gym-bro lore, and most of the older cognitive trials. The problem is that 5g/day may not meaningfully increase brain creatine. The blood-brain barrier transport system (SLC6A8) is rate-limited, and the brain’s creatine pool turns over slowly.

The CABA Alzheimer’s trial used 20g/day and achieved an 11% increase in brain creatine on MRI (5). The Gordji-Nejad sleep deprivation study used 0.35g/kg as a single acute dose (roughly 25 to 30 grams for a 75-kilogram person) and saw brain changes within hours (3). Many trials using 5g/day failed to demonstrate any measurable change in brain creatine on spectroscopy. If you’re taking creatine at 5g/day and thinking you’re optimising your brain, you’re probably optimising your muscles and hoping for a cognitive spillover that the evidence suggests may never arrive.

What I do: for general maintenance, 5g/day of creatine monohydrate. Whether it moves brain creatine meaningfully over months to years is unknown, but the safety profile is excellent and it’s a reasonable floor for muscle and bone health.

For active cognitive targeting in older adults, I use 10g/day. That’s the dose range from the trials that showed memory benefits in the 66-to-76 age group.

For acute cognitive demands under sleep deprivation (a long clinic day after a late night, an international flight followed by patient calls), I take a single dose of 0.35g/kg, roughly 25 to 30 grams, several hours before the demanding period. This is based on one study with 15 participants. It’s a rational clinical extrapolation from Gordji-Nejad’s data, not established practice, and I’m comfortable with that given the safety profile.

For a brain-loading trial over 6 to 8 weeks, 20g/day is the dose most likely to actually move brain creatine stores. Split into four 5-gram doses through the day to minimise gastrointestinal side effects.

One concern patients raise: will creatine at loading doses disrupt sleep? Ben Maaoui and colleagues tested exactly this in 2025, giving 20g/day for seven days in physically active men with sleep monitored by actigraphy (16). No impairment of sleep metrics. The energising effect on the brain doesn’t translate into insomnia.

Creatine monohydrate is the form to buy. Not creatine hydrochloride, not creatine ethyl ester, not buffered creatine, not whatever the supplement industry has repackaged this month. Creatine monohydrate is the form used in virtually every positive trial, it’s the cheapest, and no other form has demonstrated superiority in any meaningful outcome. Buy from a manufacturer with third-party testing (NSF Certified for Sport, Informed Sport, or equivalent). In Australia, creatine monohydrate powder is generally regulated as a food ingredient by FSANZ. Capsules or tablets making sports performance claims fall under TGA jurisdiction. Powder from a reputable source is the simplest and most cost-effective option.

There are no RACGP or AHMAC clinical guidelines addressing creatine supplementation for cognitive health. This is entirely off-guideline territory. I document my clinical reasoning when I recommend it to patients, and I think that’s the appropriate standard for any doctor considering this.

The Alzheimer’s disease story deserves more attention because the basic science is genuinely compelling, even though the clinical evidence is still preliminary. Zheng and colleagues published a study in 2024 examining postmortem brain tissue from patients with Alzheimer’s disease, dementia with Lewy bodies, and age-matched controls (1). They found that CKB (brain creatine kinase) is expressed exclusively in astrocytes in the human cortex. In Alzheimer’s brains, CKB immunoreactivity was reduced, and the reduction correlated with amyloid plaque load, tau pathology, and Lewy body stage. Critically, the total amount of CKB protein was unchanged: the enzyme was present but had been inactivated by post-translational oxidative modifications. The energy buffer was intact in quantity but broken in function.

If the creatine kinase system fails early in the Alzheimer’s cascade, before symptoms appear, then supplementing creatine to boost the substrate pool makes mechanistic sense as a preventive strategy. Smith and Taylor (the same group behind the CABA pilot) published a review in 2023 laying out this rationale: impaired brain energy metabolism is a hallmark of Alzheimer’s that precedes clinical symptoms, and the creatine system is specifically disrupted (18). I find this line of reasoning persuasive, and I should be transparent that it remains a hypothesis supported by postmortem studies, animal models, and one uncontrolled pilot. We do not yet have a randomised trial showing that creatine supplementation prevents or slows Alzheimer’s disease. The larger controlled CABA trial is being planned. Until it reports, enthusiasm should be proportionate to the evidence.

I’ve taken creatine for years and have increased my dose from 5g to 10g per day based on the evidence I’ve reviewed for this article. Before a day where I know I’ll be sleep-compromised and cognitively loaded, I’ll take a single higher dose of roughly 0.35g/kg, the same acute protocol Gordji-Nejad used, applied to a clinical day rather than a laboratory session.

For my patients at My Performance Doctor, I frame it simply. Creatine monohydrate at 5g/day is a well-supported supplement for muscle health, bone density, and physical performance. The safety data across decades of use shows no clinically relevant kidney or liver effects in healthy adults, even at higher doses (17). If you’re already taking it for your training, keep going.

If you’re over 60, or in that 45-to-65 window where sleep is compromised, cognitive demands are high, and you’re starting to feel the edges of age-related cognitive slowing, there’s a reasonable case for increasing to 10g/day. The evidence isn’t definitive. It’s directional. The risk-to-benefit ratio is favourable given the safety profile.

If you’re vegetarian or vegan, the case is stronger. You’re starting from a lower baseline, and the original positive trials were conducted in your population.

What creatine is not: a substitute for sleep, exercise, blood pressure control, or any of the interventions with stronger long-term evidence for brain health. It’s an adjunct. A useful one, I believe, but an adjunct. Base first, then we can get fancy.

The field is moving fast. The controlled CABA trial will be the most important data point in the next few years. I’ll update my recommendations when it reports.

For now, creatine is one of the few supplements where the basic science, the preliminary clinical data, and the safety profile all point in the same direction. I take it. I recommend it. And I’ll keep reading the evidence as it comes in, because that’s how this works.