Does the 2026 creatine RCT show creatine builds muscle and improves memory in middle age?

---

The short answer

A 12-week randomised trial from Texas A&M, published in October 2026, gave 64 middle-aged and older adults 10 g/day of creatine monohydrate or a maltodextrin placebo (PMID 42578920). The headline result is real and worth knowing: in the group that both exercised and supplemented, lean tissue mass rose 1.69 kg (95% CI 1.0 to 2.4) at six weeks and 1.27 kg (0.56 to 1.98) at twelve, against essentially no change in either placebo group.

Three things about the study are less likely to survive the trip into a headline, and all three come from the paper itself.

Participants chose whether to exercise. Randomisation applied to creatine versus placebo, not to exercise versus no exercise.

The authors call the cognitive findings exploratory. Their word, in the abstract.

The senior author chairs the scientific advisory board of a creatine manufacturer, co-founded the society that publishes the journal, and the paper appears in that journal's supplement issue. All of it is disclosed in the paper.

None of that makes the lean-mass result wrong. It makes it a result that needs reading carefully rather than repeating.

Evidence tier: 1 for the randomised supplement comparison — a registered, double-blind, placebo-controlled trial (ISRCTN83081058), read in full. Tier 3 for anything comparing exercisers to non-exercisers, because that allocation was not randomised.

Key points:

• Lean tissue mass rose 1.27 kg at 12 weeks in the exercise + creatine group, with a confidence interval excluding zero.
• Exercise participation was self-selected, so exercise-versus-no-exercise comparisons are observational.
• Dose was 10 g/day — two 5 g servings, double the usual maintenance dose.
• Roughly 12 to 20 people per cell, which the authors justify from prior work but which is small for cognitive endpoints.
• The disclosure statement is substantial and is the part nobody will quote.

What the trial did

Seventy-three sedentary adults enrolled; 64 finished and were analysed. They were aged 45 to 65 (mean 54.5), averaged 84.5 kg, and 40 of the 64 were women — a better sex balance than most supplement trials manage.

Each participant decided whether to join a non-exercise arm or an exercise-plus-weight-loss-diet arm. Within whichever arm they chose, they were then randomly assigned, double-blind, to 2 × 5 g/day of creatine monohydrate or an identical-looking maltodextrin placebo. DXA body composition, fasting bloods and a battery of cognitive tests were taken at baseline, six weeks and twelve.

That produces four groups: no-exercise placebo, no-exercise creatine, exercise placebo, exercise creatine.

The result that holds up

The lean-tissue-mass finding is the strongest thing in the paper.

Gaining lean mass during a weight-loss diet is the interesting part. People losing weight normally lose some lean tissue alongside fat, and in middle-aged and older adults that matters more than it does at 25. Both placebo groups sat flat. Only the creatine-plus-exercise group moved, and its confidence intervals do not cross zero at either timepoint.

A reporting note: the twelve-week figure printed for the no-exercise placebo group is "−0.16 kg [0.56, 1.98]". A confidence interval of 0.56 to 1.98 cannot contain a point estimate of −0.16, and those bounds are identical to the exercise-plus-creatine interval on the same line. This looks like a transcription slip in the manuscript rather than anything substantive, but we are flagging it because we read it rather than skimmed it.

The design detail that changes how you read it

From the paper's own limitations: "exercise participation was self-selected, whereas supplement assignment within each exercise-status stratum was randomized and double-blind."

That is the honest description, and the authors deserve credit for stating it plainly. It also means the study is two different grades of evidence stitched together.

The creatine-versus-placebo comparison inside each arm is a proper randomised comparison. That part is solid.

The comparison between arms is not. People who volunteer for a twelve-week exercise and dieting programme differ from people who decline one — in motivation, baseline fitness, sleep, diet quality, and a dozen things nobody measured. Any statement of the form "exercise plus creatine beat no exercise" is an observational claim wearing a randomised trial's clothes.

The abstract's phrasing — "selected whether to participate… and were then randomized, in a double-blind and counterbalanced manner" — is accurate, but it reads at speed as though the whole design were randomised. It wasn't.

Why we are not leading with the memory results

The abstract reports "favorable changes in… selected markers of cognitive function and memory" in both creatine groups. It also contains this sentence: "Individual cognitive and biomarker findings should be interpreted as exploratory."

That is the authors telling readers not to headline the cognition results. It is the right call, for a reason visible in their own methods: they ran a battery of cognitive tests and a panel of blood markers across three timepoints and four groups. Run enough comparisons and some will clear significance by chance alone. "Selected markers" is doing a lot of work in that sentence — it means some moved and some didn't.

The authors also explain their sample size: prior cognitive-nutrition work suggested 12 to 20 participants per group, and earlier creatine-cognition trials mostly enrolled 8 to 20. That is a reasonable justification for a pilot-scale signal. It is not a basis for concluding creatine improves memory in middle age.

The disclosure

Published in the Journal of the International Society of Sports Nutrition, volume 23, supplement 1.

The senior author's conflict-of-interest statement reads, in part: he "serves as Chair of the Scientific Advisory Board for AlzChem (a company that makes creatine monohydrate)", is "a co-founder of the non-profit International Society of Sports Nutrition (ISSN)", has "conducted industry-sponsored research on creatine", and sits on two further nutrition advisory boards. The other twelve authors declare no conflicts.

So: a positive creatine trial, from a lab led by the chair of a creatine manufacturer's scientific advisory board, in the journal of a society he co-founded, in a supplement issue.

This is disclosed, and disclosure is the system working. Creatine is also one of the most-studied supplements in existence, and this group has produced a great deal of the good evidence on it — expertise and industry ties tend to travel together in nutrition science, which is a structural problem rather than a personal one. The lean-mass result does not become false because of who ran it.

But a reader deciding whether to spend money on creatine is entitled to know that the trial, the journal and the advisory board share an author. Essentially no coverage of this paper will mention it, which is precisely why we are.

Where this sits in the wider creatine picture

Creatine's evidence base is unusually mixed depending on which question you ask.

For lean mass and strength in older adults, which is what this trial tested, the direction of evidence is reasonably consistent and this study adds to it.

For cancer, the picture is genuinely contested, and we have covered it separately. A 2026 UCLA paper reported creatine uptake promoting dendritic-cell activation and anti-tumour immunity in mice (PMID 42006288) — though the viral version of that study asserted a causal experiment the paper does not contain. Two independent groups report the opposite direction: creatine promoting metastasis via Smad2/3 (PMID 33811821) and via megakaryocyte CKB-STAT5B, with hyperactive platelets demonstrated in healthy human volunteers (PMID 42449108). The only randomised trial in cancer patients, N02C4, enrolled 263 people and was null on its primary endpoint and on survival (PMID 28475678).

One practical interaction worth knowing regardless: creatine raises serum creatinine because creatinine is its breakdown product. It is an artefact, not kidney injury — but several chemotherapy drugs are dosed on renal function, so anyone on creatine receiving renally dosed treatment should tell their oncology team.

What we could not check

• Per-group numbers for each outcome. The paper reports roughly 12 to 20 per cell as a design target; we did not extract completed-case counts for every endpoint.
• Which cognitive tests moved and which didn't. The abstract says "selected markers" and the authors label these exploratory; we have not reproduced the full battery here.
• Whether AlzChem supplied the creatine. The disclosure names the advisory-board role but we did not locate a statement about product donation.

Limitations

This is educational content, not medical advice.

• One trial does not settle a question, and this one is pilot-scale on its cognitive endpoints by the authors' own description.
• Self-selected exercise allocation means anything comparing exercisers to non-exercisers here is observational.
• The participants were sedentary 45-to-65-year-olds. Results do not transfer automatically to younger, trained, or older populations.
• We have not independently verified the DXA or cognitive data, only read the published report.
• Creatine is not a peptide. It sits on this site because our readers keep asking about it and because the evidence around it is routinely misreported in both directions.

The bottom line

The lean-mass finding is the real one: 1.27 kg more lean tissue at twelve weeks, during a weight-loss diet, with a confidence interval that excludes zero. For a middle-aged person dieting and lifting, that is a meaningful result from a cheap, well-characterised supplement.

The memory findings are exploratory and the authors say so. The exercise comparison is not randomised and the authors say that too. And the disclosure statement — chair of a creatine manufacturer's advisory board, co-founder of the publishing society, supplement issue — is material context that the study's own authors provided and that almost nobody else will repeat.

Read the paper rather than the headline. On this one, the authors were more careful than the coverage will be.