Brain health supplements: what the evidence says about what works

Nearly one in four adults over 50 takes a supplement marketed for brain health. The industry is built on promises of sharper memory, faster thinking, and protection against age-related decline. The evidence behind those claims is thinner than the price tags suggest.

“There’s no evidence to suggest there’s an ingredient in supplements that can improve brain health,” said Pieter Cohen, an associate professor of medicine at Harvard Medical School. The FDA does not test or approve these products before they reach store shelves.

Harvard Health put it bluntly in early 2026: “Don’t buy into brain health supplements.” A nationally representative survey found that roughly one in four older adults takes at least one such product, often believing it will protect against memory loss or dementia. Manufacturers can claim their products aid “mental alertness” and “memory support” without submitting any evidence to the FDA.

What the multivitamin studies found

The exception in the data is ordinary. The COSMOS trial, a large randomized controlled trial funded by the National Institutes of Health and led by researchers at Harvard and Mass General Brigham, found that adults 60 and older who took a daily multivitamin for about two years scored better on episodic memory tests than the placebo group. The effect was roughly equal to slowing cognitive aging by about two years.

Participants who were biologically older than their actual age at the start of the trial saw the biggest improvements. The results were published in the American Journal of Clinical Nutrition. The COSMOS researchers also tested cocoa extract and found no significant cognitive benefit.

Creatine has the best cognitive data

Of individual supplements marketed for brain function, creatine has the most consistent research. A February 2026 systematic review in Nutrition Reviews found creatine “may be favorable for cognitive function, particularly memory and processing speed” in older adults.

It works the same way in the brain as in muscle tissue: replenishing ATP energy stores. Brain creatine declines with age and stress. People with lower baseline levels, including vegetarians, older adults, and women in perimenopause, tend to see the largest gains.

A 2023 randomized controlled trial in BMC Medicine found that creatine improved cognitive performance compared to placebo. The effect sizes were modest but consistent across multiple domains.

Omega-3s and the brain: the evidence gets more complicated

The evidence on fish oil for brain health keeps pulling in different directions. A 2026 study in the Journal of Prevention of Alzheimer’s Disease tracked more than 800 older adults from the Alzheimer’s Disease Neuroimaging Initiative. Omega-3 supplement users showed faster cognitive decline than non-users on the MMSE and ADAS-Cog13 assessments. The decline was not linked to amyloid plaques or tau buildup but appeared to involve changes in synaptic function.

“These findings challenge the prevailing view of omega-3 as uniformly beneficial,” the authors wrote.

The study was observational and does not prove causation. A 2025 systematic review found that low doses may still help, but daily doses above 1,500 mg may reverse any advantages.

For dietary sources the picture is clearer. A January 2025 systematic review of 88 studies in Nutrients found that adherence to a Mediterranean diet, which includes fatty fish, was linked to slower cognitive decline.

Ginkgo biloba and B vitamins: not supported by the trials

Ginkgo biloba has decades of marketing behind it but the trial data does not cooperate. The GEM study recruited over 3,000 older adults. Average age: 79. Participants took either 120 mg of ginkgo or a placebo twice daily for nearly six years. Ginkgo did not lower the rate of developing dementia.

B vitamins play essential roles in brain function. Deficiencies cause cognitive problems. But for people who are not deficient, the clinical trials have not shown consistent cognitive benefits from supplementation.

Lion’s mane and nootropics: promising but unproven

Lion’s mane contains hericenones and erinacines, compounds that may stimulate nerve growth factor. A September 2025 systematic review in Frontiers in Nutrition found promise for mood and neuroprotection but said more human trials are needed.

Proprietary nootropic blends are harder to evaluate. They combine ingredients in undisclosed ratios at doses no one has studied. Their effects in healthy adults are largely unknown.

What longevity experts actually take

Peter Attia, a longevity-focused physician, has published his supplement protocol. For brain health he takes Theracurmin, a high-absorption form of curcumin. He cites an 18-month study of 40 people in which 180 mg per day improved memory and attention scores and reduced amyloid and tau accumulation on brain scans. His team rates the evidence at 6 out of 10.

He takes magnesium L-threonate before bed. The form is designed to cross the blood-brain barrier. His team rates it at 2 out of 10. He skips multivitamins, preferring targeted supplements chosen by regular blood work.

The bottom line on brain supplements

Diet comes first. A Mediterranean or MIND dietary pattern delivers omega-3s, polyphenols, B vitamins, and antioxidants from food in combinations pills cannot match. A 2022 cohort study in Neurology of 72,000 UK Biobank adults found that higher ultra-processed food intake was tied to higher dementia risk.

For people who want to add supplements to a solid diet, three options have the best evidence. A daily multivitamin for older adults, backed by the COSMOS trial. Creatine for memory and processing speed, especially at lower baseline levels. Omega-3s from fatty fish, where evidence is strong, but the supplement form is less reliable and may carry risks at high doses.

“Don’t buy into brain health supplements” was the blunt title of the Harvard Health review. The evidence is not that absolute. The multivitamin and creatine data are real. But most products in this category sell on promises the research does not back.