Ginkgo Biloba: Two Huge Prevention Trials, Two Failures

The claim

Ginkgo is the most recognized name in this category. It has been sold as a memory supplement for decades, it appears in a large share of multi-ingredient “brain” formulas, and its consumer reputation is essentially fixed: it is the herb for memory, the one your relatives have heard of, the safe default. That reputation was built before the evidence arrived, and the evidence, when it arrived, was not kind.

The specific claim worth testing is the strong one — that taking ginkgo protects an aging brain, lowering the chance of developing dementia or slowing the slide toward it. It is the reason people take it for years rather than weeks. It is also the claim that has been directly tested at a scale almost nothing else in this field has matched.

Where it came from

Unlike most of what gets sold as a nootropic, ginkgo’s story starts with real pharmaceutical development. EGb 761 is a standardized extract — 24% flavone glycosides, 6% terpene lactones — sold in Europe under names including Tebonin, Tanakan and Rökan, and used in nearly all of the quality research. That standardization matters: it means the trials tested a defined preparation rather than whatever a given bottle of leaf powder happens to contain.

Early work in diagnosed dementia produced enough signal to justify the obvious next question. If the extract helps people who already have dementia, does giving it to people who do not yet have dementia prevent it? Two governments funded the trials to find out. That is the correct sequence, it is expensive, and it almost never happens in this industry.

What the evidence actually shows

Trial	Population	Dose	Duration	Outcome
GEM (Ginkgo Evaluation of Memory), DeKosky et al., JAMA 2008	≈3,069 older US adults	240 mg/day EGb 761	Median 6.1 years	No reduction in incidence of all-cause dementia or Alzheimer’s disease1
GuidAge, Vellas et al., Lancet Neurology 2012	≈2,854 older French adults with memory complaints	Standardized ginkgo extract	5 years	No reduction in progression to Alzheimer’s disease2
Cochrane review 2026 (Wieland et al., CD013661.pub2), searched to Nov 2024	82 studies, 10,613 participants, 52 published in Chinese	Various	Up to 12 months	Dementia: possible modest 6-month improvement in global condition, cognition and daily activities, with considerable variation. MCI: “probably makes little or no difference at six months.” Subjective memory complaints: inconclusive34

Take the scale seriously for a moment. Between them, GEM and GuidAge randomized nearly 6,000 people and followed them for five to six years. That is more participant-years than the entire trial literature behind most of the ingredients in a typical brain supplement, combined. These were not underpowered pilot studies whose null results could be dismissed as a failure to detect a small effect. They were built to find a prevention benefit, and there was none to find.

The Cochrane review then closes the adjacent door. In mild cognitive impairment — the stage people most want to intervene at — ginkgo probably makes little or no difference at six months. For MS-related cognitive problems, no meaningful improvement at three months. For subjective memory complaints, the kind that send a healthy 55-year-old to a supplement aisle, the evidence is inconclusive.

The one population where ginkgo did something

Honesty runs both ways, and the finding that most articles about ginkgo omit is the positive one. In people with diagnosed dementia, over about six months, at 240 mg/day of EGb 761, the Cochrane review found possible modest improvements in global condition, cognition and activities of daily living compared with placebo — with results that varied considerably between studies.³ That is a real finding in a real population, and it is why the tier for ginkgo is not simply “negative.”

What that finding does not license is the leap most marketing makes. An effect in people with a diagnosis, over six months, on clinician-rated global scales, is not evidence of an effect in a healthy 45-year-old who forgets names at meetings. Populations do not transfer. This is the single most common error in supplement writing, and it is the reason we attach a population to every effect claim on this site.

Why the claim persists

• Name recognition outlives evidence. Ginkgo entered public consciousness in the 1990s. GEM reported in 2008 and GuidAge in 2012, into a market that had already made up its mind.
• Null results get reported once. A failed prevention trial is a single news cycle. A supplement is on a shelf for twenty years.
• The dementia signal is quoted without its population. “Cochrane found benefits for cognition” is a true sentence that becomes false the moment you drop the words “in people with diagnosed dementia, at six months, with considerable variation.”
• It is cheap, familiar and feels harmless. The last of those is the part worth examining.

Ginkgo is also one of the most common ingredients in multi-ingredient blends, where its recognizability does the persuading and its dose is rarely disclosed. Non-standardized ginkgo products are not comparable to EGb 761, and product-quality and adulteration problems are documented in this category. A capsule containing “ginkgo biloba” with no extract standardization and no stated dose has no relationship to the material that was tested in either trial.

The safety side, which is not nothing

Across trials up to 12 months, Cochrane found probably little or no difference in overall or serious adverse events versus placebo, and that is a fair summary of the average case.³ The interaction profile is where the real risk sits, and it is not distributed evenly across readers.

Put those two facts together and the risk-benefit arithmetic for a healthy person is straightforward. The expected benefit for prevention is zero, on the best evidence anyone has. The expected harm is small but not zero, and it concentrates in exactly the population most likely to be taking a daily supplement for their brain: older adults, who are also the people most likely to be on an anticoagulant or an antiplatelet drug.

What to do instead

The honest alternative is duller than ginkgo and better supported. In the COSMOS trial programme, a plain daily multivitamin-mineral produced the only large, randomized, replicated positive cognitive result in this whole field: COSMOS-Mind (2,262 participants, three years) found improved global cognition versus placebo, estimated as slowing cognitive aging by roughly 60%, about 1.8 years over three years; COSMOS-Web (3,500+ participants) found memory improvement at one year sustained through three; a meta-analysis across the COSMOS cognitive studies supported a modest overall benefit.⁶⁷⁸ Effect sizes are small, the participants were largely well-educated older US adults, and the mechanism is not established. It is still more than any botanical in this category can show.

Creatine monohydrate is the other compound with a defensible file, and its limits are as instructive as its results: memory effects around SMD 0.29 to 0.31, concentrated in older adults, vegetarians and people under metabolic stress, with essentially nothing in young, well-fed, well-rested people — and EFSA rejected a creatine-and-cognition health claim in 2024 on the grounds that a cause-and-effect relationship was not established.⁹¹⁰¹¹

If your reason for taking ginkgo is a specific worry about your memory rather than a general wish to do something, the right move is not a different capsule. It is a clinician who can look at your medication list, your sleep, your thyroid and your B12 status. Our page on brain fog covers the situational causes; the tier list ranks what has human trials behind it; and the case against brain supplements makes the broader argument, written by a site that reviews them.

What would change our mind

A third prevention trial of similar size and duration to GEM, with a positive pre-registered primary endpoint, would change this page — and given the two failures behind it, it would need to be at least as large and at least as long. Something narrower could also move us: a well-powered trial in a defined subgroup, published with its protocol, showing benefit where GEM and GuidAge saw none. What will not move us is another set of small short trials in mixed populations, which is what the existing literature is already mostly made of.

Frequently asked questions

Does ginkgo improve memory in healthy people?

There is no good evidence that it does. Cochrane rates the evidence for subjective memory complaints as inconclusive, and the two large prevention trials in older adults found no benefit on their primary endpoints.¹²³ “No evidence of benefit” is the answer, and it is a real answer rather than a dodge.

How much ginkgo did the trials use?

120 to 240 mg/day of EGb 761, usually split into two doses. GEM and the more positive dementia work used 240 mg/day. If a product does not say EGb 761 or give an equivalent standardization — 24% flavone glycosides, 6% terpene lactones — you cannot map its contents onto any of that research.

My parent has dementia. Should they take it?

That is a question for their clinician, and it is a genuine question rather than a rhetorical one, because the dementia signal is the one place the evidence is not null. The reason it needs a clinician is the medication list: bleeding risk with anticoagulants and antiplatelets is the specific concern, and it is common in exactly this group.

Isn’t ginkgo good for circulation?

Circulation is a mechanism, not an outcome. The mechanism is real enough to be the basis of the bleeding-risk warning. It did not translate into dementia prevention across nearly 6,000 people and six years, which is the only test that matters for the claim people buy it on.

Is the ginkgo in my multi-ingredient formula the same thing?

Almost certainly not. Non-standardized ginkgo is not comparable to EGb 761, blends rarely disclose per-ingredient doses, and quality and adulteration problems are documented in this category. A blend containing an unstated amount of unstandardized ginkgo has not been tested by anyone.

Related reading

• Do Nootropics Actually Work? An Honest Tier List
• The Case Against Brain Supplements
• Nine Ingredients That Signal a Bad Brain Supplement
• Prevagen and the Apoaequorin Verdict
• Vinpocetine: FDA Says It’s Not a Dietary Ingredient
• How We Rate Evidence