The memory supplement guide: what the evidence actually shows

In 2008 the Ginkgo Evaluation of Memory study published its main result. It had randomised 3,069 adults aged 75 and over to 120 mg of ginkgo biloba twice a day or placebo, and followed them for a median of just over six years, looking for dementia. It found none: the extract did not reduce the overall incidence of dementia or of Alzheimer's disease, either in people with normal cognition or in those with mild cognitive impairment.

That result matters more than any single supplement finding, because ginkgo was the category's flagship. It had plausibility, centuries of traditional use, positive small trials and a standardised extract with a known chemistry. It was tested properly, at a real dose, for long enough, in the right population, and it did not work for the outcome that matters most.

A reader should take two things from it. First, that this category can be tested, which means claims in it are answerable rather than merely arguable. Second, that plausibility is not evidence, and that most products on the shelf have less behind them than the one that failed.

Walk along a shelf and you will find perhaps eight recurring ingredients. It is worth knowing where each one's evidence sits, because the marketing does not distinguish between them.

Ginkgo biloba has the largest trial and it was negative for dementia incidence. B vitamins have a real biological story: severe deficiency of niacin causes pellagra, whose third classical symptom is dementia, and an observational cohort of 3,718 older residents of Chicago found higher dietary niacin intake associated with lower Alzheimer's incidence and slower decline. That is a cohort about food rather than a trial of a capsule.

Amino acids feeding nitric oxide — arginine and citrulline — have good evidence for blood pressure and thin evidence for cognition. A 2025 meta-analysis found citrulline lowered blood pressure in middle-aged and older adults; one 2025 trial measured cerebrovascular function at 8.8 g a day, in athletes.

Beta-alanine and carnosine have two small and genuinely interesting results: 2.4 g a day of beta-alanine changed white-matter measures in the hippocampus and amygdala of adults aged 60 to 80, and 2 g a day of carnosine improved cognitive outcomes in the younger participants of one trial. Small, unreplicated, and more than most ingredients have.

Most readers of a guide like this already have a bottle in a cupboard. Three practical thoughts, in descending order of usefulness.

Take it consistently or stop. A capsule taken four mornings out of seven is a test of nothing, and the outcome most people describe at the end of a half-hearted month is the outcome of the half-heartedness.

Count something rather than feel something. Memory is the one outcome where the instrument doing the measuring is the thing being measured, so a diary beats a recollection by a wide margin.

And do the free things alongside it, not instead of it. If the sleep improves at the same time, the sleep has the better evidence, and the right conclusion at the end is a cheerful one: something worked, and the cheapest candidate gets the credit.

• A large randomised trial of a combination product, at printed amounts, with a cognitive endpoint. Nothing like it currently exists.
• Replication of the beta-alanine imaging result in a larger sample.
• A citrulline trial in ordinary adults at rest rather than athletes mid-session.
• Any supplement trial that measures what buyers actually care about — names, appointments, where the keys are — rather than a battery of laboratory tasks.

Until then, the summary a reader can rely on is short. Treat your hearing. Treat your blood pressure. Sleep. Consider a multivitamin, which is cheap and has one positive randomised result. Everything past that point is a reasonable experiment on yourself, and should be bought as one.