Protocol
Creatine 3–5 g/day
Memory improved overall in a 2023 meta-analysis A study that pools the numbers from several trials into one combined estimate. of eight randomized trials, but the effect sits in older adults and is absent in younger ones, heterogeneity How much the results of the pooled trials differ from each other. High heterogeneity makes a combined estimate less reliable. is substantial, and the positive older-adult trial in that pool used 20 g/day for one week, not 3 to 5 g.
Grade C: human trials exist, but they are small, mixed or not yet repeated. This is not a finding that it does not work. How to read a grade →
Compare this record with another, side by side →
How this letter has changedLatest: October 2026 · Corrected.
The record cited Candow 2015 (PMID PubMed ID: the permanent number the US National Library of Medicine gives a published paper. Each one links to that paper's record on PubMed. 25993883) for creatine plus training on cognition; that trial measured lean mass and strength only and is now labeled as such. Avgerinos 2018 (PMID 29704637) is a systematic review A review that searches for every study on a question using a method set in advance, then assesses them together. of six trials without pooling, not a trial, and is relabeled. Prokopidis 2023 (PMID 35984306) now carries its own declaration, an author on the advisory board of a creatine manufacturer and another who advised a creatine seller, and the letter by Eckert and Pascher 2023 (PMID 36644917) alleging double counting, with the authors' reply (PMID 36644912). Added at the record's own dose: Sandkühler 2023 (PMID 37968687), the largest preregistered trial, null on both primary tests with side effects more frequent on creatine (RR 4.25), now a safety note; Alves 2013 (PMID 24098469), 24 weeks in older women, null. Rawson 2008 (PMID 18579168) is null in young adults. McMorris 2007 (PMID 17828627), the older-adult positive trial, used 20 g/day for one week, so the description no longer implies the older-adult effect was shown at 3 to 5 g. Xu 2024 (PMID 39070254) is added as the later meta-analysis, pooling patients with healthy adults. Letter unchanged.PMID 25993883 PMID 29704637 PMID 35984306 PMID 36644917 PMID 36644912 PMID 37968687 PMID 24098469 PMID 18579168 PMID 17828627 PMID 39070254
The cognitive protocol carried a Grade A resting on Prokopidis 2023 (PMID 35984306). Rereading the meta-analysis against the letter: the pooled memory effect is SMD 0.29 (95 percent CI 0.04 to 0.53) with I-squared of 66 percent, concentrated in adults aged 66 to 76 (SMD 0.88) and absent in adults aged 11 to 31 (SMD 0.03, -0.14 to 0.20). A signal that reverses to nothing across half the age range is mixed trial data. Regraded C. This was found while splitting the creatine supplement record by use, which had carried one letter across muscle strength, cognition and brain energy reserve. Creatine's Grade A is scoped to measured strength, where it rests on instrument-recorded one-repetition maximum in two mixed-sex meta-analyses, and is unchanged. The bone density use is newly graded D by Route 2 Human randomized trials have been run and did not show benefit on the stated outcome. One of the two routes to the lowest letter on EBL's scale. on two 2-year randomized trials that found no effect.PMID 35984306
Every change to a letter on this site is dated and published. The full evidence log →
This protocol is creatine at 3 to 5 grams a day for memory and thinking. A review of eight studies found memory improved overall. The gain showed up in adults aged 66 to 76, not in younger adults. The studies differed a lot, and some used doses up to 20 grams a day. So the memory benefit at this dose is not settled.
Memory improved overall in a 2023 meta-analysis of eight randomized trials, but the effect sits in older adults and is absent in younger ones, heterogeneity is substantial, and the positive older-adult trial in that pool used 20 g/day for one week, not 3 to 5 g. The largest trial at 5 g/day, preregistered, was null on both primary tests, and a 24-week trial at 5 g/day in older women found no cognitive change. Monohydrate form. Any timing.
Why this gradeGrade C, small or unreplicated trials. The reasoning, and what the grade does not cover.
Grade C on objectively measured memory and cognition. A meta-analysis of eight randomized trials found memory improved overall, but the effect sat in adults aged 66 to 76 and was absent in younger adults, heterogeneity was substantial, and doses ranged up to 20 g/day (Prokopidis et al., 2023; PMID 35984306); a published letter argued the analysis double-counted (Eckert and Pascher, 2023; PMID 36644917). The largest preregistered trial at 5 g/day was null on both primary tests (PMID 37968687), and a 24-week trial at 5 g/day in older women found no cognitive change (PMID 24098469).
WHAT THIS GRADE DOES NOT COVERmuscle strength and size, graded separately on the creatine record; adults under 66, in whom the pooled memory effect was absent; and doses above 5 g/day. The C is for cognition at 3 to 5 g/day only.
A grade describes the human evidence for a named use. It is not advice, and it is not a judgment on any individual prescription. How the letters are decided →
No prominent contraindication A condition or medication that makes a treatment unsafe or unsuitable for a particular person. at 3 to 5 g/day. In the largest cognition trial at 5 g/day, Sandkühler 2023 (PMID 37968687), side effects were reported significantly more often on creatine than on placebo An inactive substitute made to look like the real treatment, so participants cannot tell which one they received. (risk ratio 4.25); the trial does not list which. Serum creatinine rises with creatine intake and is not by itself a sign of kidney injury.
The evidence, in fullTen papers, each linked to its PubMed record.
Prokopidis et al. (Nutrition Reviews, 2023; PMID 35984306): meta-analysis of eight randomized trials, memory SMD 0.29 (95 percent CI 0.04 to 0.53), I-squared 66 percent, with the effect concentrated in adults aged 66 to 76 (SMD 0.88) and absent in those aged 11 to 31 (SMD 0.03, -0.14 to 0.20); doses in the pool ranged from about 2.2 to 20 g/day. The paper declares that one author sits on the scientific advisory board of a company that manufactures creatine, has run industry-sponsored creatine research and received creatine donations, and that another author previously advised a company that sold creatine. Eckert and Pascher (Nutr Rev, 2023; PMID 36644917) wrote that the analysis double-counted data in a way that could produce false-positive findings; the authors replied (PMID 36644912). Xu et al. (Front Nutr, 2024; PMID 39070254): meta-analysis of 16 randomized trials, 492 participants aged 21 to 76, memory SMD 0.31, pooling healthy adults with patients and finding larger effects in people with disease. Null trials at this record's dose: Sandkühler et al. (BMC Med, 2023; PMID 37968687): preregistered crossover A trial design in which each participant receives both the treatment and the comparison, one after the other, and serves as their own control. in 123 adults, 5 g/day for 6 weeks, no significant effect on either primary test (Raven's matrices p = 0.33, backward digit span p = 0.064), a small Bayesian signal only, and side effects reported significantly more often on creatine than placebo (RR 4.25). Alves et al. (PLoS One, 2013; PMID 24098469): 24 weeks, 5 g/day after a loading week, older women, no change on any cognitive measure. Rawson et al. (Physiol Behav, 2008; PMID 18579168): 22 young adults, 6 weeks, null on a full cognitive battery. The positive older-adult trial: McMorris et al. (Aging Neuropsychol Cogn, 2007; PMID 17828627): 32 older adults, 5 g four times a day for one week, improved on all tasks but one; 20 g/day, not this record's dose. Avgerinos et al. (Experimental Gerontology, 2018; PMID 29704637): systematic review of six randomized trials in healthy individuals with no pooled analysis, finding short-term memory and reasoning may improve and other domains conflicting; not a trial. Candow et al. (Appl Physiol Nutr Metab, 2015; PMID 25993883): 32-week trial of creatine timing around resistance training in older adults measuring lean mass and strength only; it has no cognitive endpoint The outcome a trial measures to decide whether the treatment worked, such as heart attacks, blood pressure or a symptom score. and is not evidence for this record. Grade C: a pooled signal that reverses to nothing across half the age range, with the largest trial at the record's dose null, is mixed trial data, not the consistent signal a higher letter needs. Creatine's Grade A is scoped to measured strength and is unaffected.
Grip strength or dynamometer output at baseline and every 8 weeks; in older adults, lean mass retention on DEXA scan at 12 and 24 weeks is the most clinically relevant endpoint. Subjective fatigue and exercise recovery scores are secondary. Serum creatinine may rise slightly: this is not a sign of kidney stress but reflects increased creatine turnover and should be distinguished from true GFR change.
Evidence notes: meta-analysis, randomized trial
Endpoint graded: Objectively measured memory and cognitive test performance (functional)
Last reviewed: July 2026
Educational use only · Not medical advice · EBL grades published human evidence and does not recommend treatment. Stopping a prescription is a decision for your physician.
Evidence Based Longevity, Inc. is owner-operated and edited by Arnold Sandlow, D.C., a California-licensed chiropractor. Evidence Based Longevity publishes in English only · Machine translations are not produced or reviewed by EBL for accuracy
© Evidence Based Longevity, Inc.. All rights reserved
These statements have not been evaluated by the Food and Drug Administration. Any grade on this page describes the strength of published human trial evidence in the populations studied. It is not a recommendation for any individual and is not medical advice. Starting, stopping or changing any supplement or medication is a decision for your physician.
How EBL searched for the trials (1 search)
The search that was run, what it returned, and what was left out and why, so the absence of evidence can be checked as well as its presence.
creatine[tiab] AND (cognit*[tiab] OR memory[tiab]) AND (randomized controlled trial[pt] OR meta-analysis[pt])Grades move when new trials publish. A free monthly email lists every grade EBL moved, with the reason and the trial behind it: the sign-up is on the Evidence Log →