Creatine is now sold for the brain as much as for the bench press, and the marketing runs four distinct claims: that it helps in dementia, that it clears brain fog, that it helps ADHD, and that it improves brain function generally. These are not one claim with four phrasings. They rest on four different bodies of evidence, and two of those bodies are empty.
Establishing that took an unusual amount of care, for a reason worth stating before any of the numbers.
Why every count you will see is inflated
In magnetic resonance spectroscopy — the scan used to measure chemistry inside a living brain — the creatine signal is the standard internal reference. Results are reported as ratios against it: glutamate-to-creatine, N-acetyl-aspartate-to-creatine, choline-to-creatine. The word "creatine" therefore appears in an enormous number of neuroscience abstracts in which no one took any creatine.
How enormous: 5,730 PubMed records pair creatine with spectroscopy or MRS.
The distortion is not theoretical. A search for creatine with ADHD returns 77 records and exactly one randomised controlled trial — and that trial administered methylphenidate for twelve weeks to 21 children in Indonesia, measuring Glu/Cr, NAA/Cr, Cho/Cr and mI/Cr ratios in the prefrontal cortex (PMID 22318191). Creatine is the denominator. Anyone quoting "there is randomised trial evidence on creatine and ADHD" has read a count, not a paper.
So the counts below come from the trial registry, where an intervention is a field rather than a word in a sentence.
The registry census
ClinicalTrials.gov, read on 19 September 2026, with creatine named as the intervention:
| Condition searched | Registered studies |
|---|---|
| All creatine interventional studies (control) | 359 |
| Cognition or memory | 27 |
| Depression | 26 |
| Dementia or Alzheimer disease | 8 |
| Parkinson disease | 2 |
| ADHD | 0 |
The control at the top is what makes the zero readable. The registry holds 359 creatine studies and 27 of them are about cognition, so the field is reachable and populated. ADHD returns nothing because nothing was registered, not because the query failed.
Claim 1 — dementia and Alzheimer's: one 20-person trial with no posted results
The eight dementia hits do not survive inspection. Seven are Huntington disease trials, which match because dementia appears in that disease's condition list — CREST-E, PRECREST and the premanifest extension studies, all testing creatine as a neuroprotectant in Huntington's. One more is an Ebola and Marburg vaccine study, matched incidentally.
That leaves exactly one Alzheimer trial: NCT05383833, "Creatine to Augment Bioenergetics in Alzheimer's", 20 participants, status completed, results posted: none.
Twenty people, and the results have not been made public. That is the entire registered evidence base for the largest claim being made.
Claim 2 — Parkinson's: tested at scale, and it failed
This is the one neurological claim that received a proper test. NET-PD Long-term Study 1 randomised 1,741 people with early Parkinson disease to creatine monohydrate 10 grams a day or placebo for a planned minimum of five years across 45 sites. A planned interim analysis found creatine no better than placebo and the trial was terminated for futility (JAMA 2015;313(6):584–93, PMID 25668262; NCT00449865).
A later analysis of the same cohort produced a finding that has never entered the consumer conversation. Among the 1,549 participants with caffeine-intake data, caffeine was not associated with progression overall — but higher caffeine intake was associated with significantly faster progression among the subjects taking creatine (Simon et al., Clin Neuropharmacol 2015;38(5):163–9, PMID 26366971). It is an observational analysis inside a randomised trial, so it shows an association rather than a cause, and it is the largest human dataset that exists on the two substances together. We trace what else is known about that pair on creatine and caffeine.
Claim 3 — ADHD: nothing, at all
Zero registrations. Zero supplement trials. The two PubMed records that survive a search for creatine, ADHD and supplementation are a scoping review of maternal nutrition and neurodevelopment (PMID 34684531) and a review of cognition-enhancing drug use, which notes only that creatine "may have mild cognition-enhancing properties" while observing that study results are confounded by exercise (PMID 26881770).
A review sentence containing "may" is not evidence about ADHD. It is a reviewer being careful about a different question.
Claim 4 — brain fog: the phrase has never been studied
A PubMed search for creatine alongside the exact phrase "brain fog" returns zero records. The control — creatine records tagged as randomised controlled trials — returns 2,854.
Brain fog is not a clinical endpoint; no instrument measures it and no trial has registered it. Every claim that creatine clears it is an extrapolation from studies of memory or of sleep deprivation in healthy people, which is a different population answering a different question.
The healthy-brain question, which is a real literature
None of the above bears on whether creatine improves cognition in healthy adults. That question has been studied repeatedly, it was formally assessed by the European Food Safety Authority in 2024, and the panel concluded that a cause-and-effect relationship has not been established. We set out that assessment, the meta-analyses behind it and the age split inside them in the cognition section of creatine benefits, and we grade cognitive claims across compounds in the cognitive claims ledger. Repeating it here would duplicate rather than add.
What would change this page
One thing, cheaply: NCT05383833 posting its results. Twenty participants will not settle anything, but it is the only registered Alzheimer trial and its silence is currently doing the work that data should do. Beyond that, an ADHD registration would move that row off zero, and a trial registering brain fog — or any validated proxy for it — would give the phrase its first measurement.
Until then, the honest statement of the position is that creatine has one large, negative neurological trial, one tiny unreported one, and two claims with no literature at all. This page reports what has been studied and what has not; decisions about any diagnosed condition belong with a clinician, per our editorial standards.
Key figures
- Registered creatine studies: 359 · cognition or memory: 27 · ADHD: 0
- Alzheimer registrations: 1, n = 20, results posted: none
- Creatine × spectroscopy records: 5,730 — the reason naive counts mislead
- Parkinson trial: 1,741 randomised, 10 g/day, terminated for futility
- Caffeine × creatine interaction analysis: 1,549 participants
- Creatine × "brain fog" records: 0, against a control of 2,854
Sources
- Writing Group for the NINDS Exploratory Trials in Parkinson Disease (NET-PD) Investigators. Effect of creatine monohydrate on clinical progression in patients with Parkinson disease: a randomized clinical trial. JAMA 2015;313(6):584–93 — PMID 25668262 · doi:10.1001/jama.2015.120
- Simon DK, Wu C, Tilley BC, et al. Caffeine and Progression of Parkinson Disease: A Deleterious Interaction With Creatine. Clin Neuropharmacol 2015;38(5):163–9 — PMID 26366971 · doi:10.1097/WNF.0000000000000102
- Effect of 12-week administration of 20-mg long-acting methylphenidate on Glu/Cr, NAA/Cr, Cho/Cr, and mI/Cr ratios in the prefrontal cortices of school-age children in Indonesia: a study using 1H magnetic resonance spectroscopy (MRS). Clin Neuropharmacol 2012 — PMID 22318191
- The Changing Drug Culture: Use and Misuse of Cognition-Enhancing Drugs. FP Essent 2016 — PMID 26881770
- Maternal Nutrition and Neurodevelopment: A Scoping Review. Nutrients 2021 — PMID 34684531
- ClinicalTrials.gov. Creatine to Augment Bioenergetics in Alzheimer's — NCT05383833
- ClinicalTrials.gov API v2 — condition census against creatine as intervention, run 2026-09-19 — clinicaltrials.gov
- National Library of Medicine. PubMed E-utilities — census searches for creatine with ADHD, dementia, brain fog and spectroscopy, run 2026-09-19 — pubmed.ncbi.nlm.nih.gov
All sources retrieved 2026-09-19. Literature and registry counts are as of that date.