Creatine: what three decades of trials established

A buffer for very short efforts

Creatine's mechanism is specified precisely enough to bound the evidence in advance. The free energy released when phosphocreatine is hydrolysed can be used as a buffer to resynthesize ATP, in the International Society of Sports Nutrition's phrasing.¹ Around 95% of the body's creatine sits in skeletal muscle, about two-thirds of it phosphorylated.¹

A larger reserve can therefore only matter where demand is brief and near-maximal: a heavy set, a sprint, a jump, the last seconds of an interval. Supplementation raises muscle stores by roughly 20–40% above baseline,¹ and the performance literature follows that mechanism rather than endurance.

Muscle and strength: the settled part

The clearest recent estimate for body composition pools 35 randomised trials and 1,192 participants.³ Across every trial, with and without exercise, creatine increased lean body mass by 0.68 kg (95% CI 0.26–1.11) — an average that conceals the result that matters: combined with resistance training the difference was 1.10 kg (0.56–1.65), regardless of age; in trials with no exercise it was 0.03 kg (−0.65 to 0.70), an interval centred on nothing.

Strength is similar in size and rougher at the edges. A 2025 meta-analysis of 69 studies and 1,937 participants reported effects lift by lift.⁴

The significant effects are small, and the non-significant ones are imprecise rather than absent: the leg press interval runs from −0.74 kg to 7.00 kg. Heterogeneity is low for the barbell lifts and higher for the vertical jump and Wingate power, where protocols vary more between labs. The same analysis found gains in younger adults but not older ones, and in men but not women; its authors call for more well-powered trials in women and older adults before those subgroup differences can be read as biology.⁴

Dose, loading, and the forms sold at a premium

The loading protocol most people have heard of comes from muscle-biopsy experiments in 31 men published in 1996.⁵ Twenty grams a day for six days raised muscle total creatine by about 20%, and the elevation held on 2 g a day for a further 30 days.

The authors concluded that the ingestion of 3 g creatine/day is in the long term likely to be as effective at raising tissue levels as this higher dose.⁵ Thirty years on, that has held: a 2021 review states that accumulating evidence indicates that you do not have to 'load' creatine, and that 3–5 g a day suffices.⁶ Loading buys about three weeks and nothing else. The position stand's version is 0.3 g per kg daily for five to seven days, then 3–5 g a day, with larger athletes possibly needing 5–10 g.¹

Among the alternatives sold at a markup — citrate, ethyl ester, nitrate, buffered, liquid — the position stand judges claims of better absorption or greater muscle uptake currently unfounded.¹ The most direct head-to-head is a double-blind trial in 36 resistance-trained participants comparing a buffered creatine against monohydrate over 28 days; neither buffered arm produced greater changes in muscle creatine content, body composition, strength or anaerobic capacity.⁷ Read that one with its funding in view: it was paid for by a manufacturer of creatine monohydrate, a conflict pointing towards the result it found.

Reading the safety file precisely

The longest observation in athletes ran 21 months: 98 Division IA college football players took creatine or non-creatine supplements, with fasting blood and 24-hour urine collected at nine time points across a 69-item panel.² No significant differences emerged. The limitation is structural and the authors state it: the study was open-label and players chose their own group. It can rule out gross toxicity; it cannot control for who self-selects into supplementing.

The position stand cites trials at 20 g a day for five days and up to 10 g a day for ten months to five years, with no effect on creatine clearance, glomerular filtration rate, tubular resorption, or glomerular membrane permeability.¹ Note the confound: blood creatinine, the number on a routine renal panel, tracks muscle mass and dietary creatine intake.⁶ A supplementing lifter can show a raised creatinine and a worse-looking estimated GFR with nothing having happened to their kidneys.

Two persistent myths do not survive the trials. On cramping and dehydration: experimental and clinical research does not validate the notion that creatine supplementation causes dehydration and muscle cramping.⁶ On hair loss, the entire basis is one study of college rugby players in which dihydrotestosterone rose 56% after loading and stayed 40% above baseline. It has never been replicated; no rise in total testosterone was found; the values stayed within normal clinical limits; and baseline DHT was 23% lower in the creatine group. No human study has reported hair loss.⁶

Where the data are thin deserves equal precision. The evidence in children and adolescents is called limited by its own reviewers.⁶ In pregnancy the human evidence is observational and the therapeutic case theoretical.¹ And the reassuring renal data come from trials that enrolled healthy people, which is exactly why they cannot speak to pre-existing kidney disease; the case reports that do exist were confounded by medications, prior kidney disease, other supplements and inappropriate doses.⁶ No signal in healthy adults is not the same claim as safe in everyone.

The frontier: cognition, and what a reanalysis did to it

Creatine is also present in the brain, and the idea that topping up a cerebral energy buffer helps thinking under load has produced a decade of small trials. A 2018 systematic review found six randomised trials covering 281 people; short-term memory and reasoning may improve, but for long-term memory, spatial memory, attention, executive function, reaction time and mental fatigue the results were conflicting, and performance stayed unchanged in young participants.⁸

A 2023 meta-analysis of eight trials in 225 participants then reported an overall benefit to memory (SMD 0.29, 95% CI 0.04–0.53, I² = 66%), driven by adults aged 66–76 (SMD 0.88, 0.22–1.55) with nothing in 11- to 31-year-olds (SMD 0.03, −0.14 to 0.20).⁹ A letter to the editor then pointed out that several memory outcomes per study had counted the same participants more than once. The authors accepted the criticism, reran the analysis on composite scores, and the headline did not survive: creatine does not improve overall memory performance (standardized mean difference [SMD], 0.19; 95% confidence interval [CI], –0.07, 0.46; I2 = 4%; P = 0.15), while the older-adult effect held at 0.80 (0.25–1.34).¹⁰ That correction is the most useful document in the file.

The sleep-deprivation finding that circulated widely is a single-dose crossover study in 15 healthy adults given 0.35 g per kg of body mass — around 25 g for an 80 kg person — during 21 hours of sustained wakefulness. Cerebral phosphate metabolites shifted, word memory improved by 10.3 ± 3.8% (p = 0.005), and processing speed rose on language, logic and numeric tasks.¹¹ Fifteen people, one night, one dose: nothing about a daily habit follows from it.

The one variable these trials never leave out is the training itself. Kettle keeps strength sessions, runs, rides, food and water on the iPhone, with data stored on the device and written to Apple Health with your permission.