Of everything I’ve looked at for this site, ginkgo has the best evidence base. Not the best results — the best evidence base. Nearly six thousand people, across two trials, followed for five and six years respectively, using a single well-characterised preparation at the studied dose.
That makes it unusually informative, and not for the reason the supplement aisle would like.
I came to it from a different direction, though. On a pharmacy shelf there are typically two ginkgo products side by side: one says “Ginkgo Biloba 120 mg,” the other says “standardised to 24% flavone glycosides, 6% terpene lactones” and costs roughly three times as much. Nothing on either box explains why the second bothers to say that.
Finding out what the difference is demolished an assumption I’d been carrying — that a plant extract occupies some comfortable middle ground between food and pharmaceuticals, and is therefore gentler and more forgiving than an isolated compound.
It doesn’t. Not remotely.
Educational, not medical advice. Ginkgo has a documented interaction with anticoagulants and antiplatelet drugs; if you take any, or have surgery scheduled, this is a conversation for your doctor before it’s a purchase.
What is actually in the capsule
Nearly every serious trial of ginkgo has used the same preparation: EGb 761, a proprietary extract made by Dr. Willmar Schwabe in Karlsruhe, Germany. Its specification is public, and worth reading slowly.
It is a dry extract of ginkgo leaves at a drug-to-extract ratio of 35 to 67 to one. The extraction solvent is 60% acetone. The result is adjusted to contain 22–27% ginkgo flavone glycosides and 5–7% terpene lactones, the latter broken down further into 2.8–3.4% ginkgolides A, B and C and 2.6–3.2% bilobalide. And ginkgolic acids — which occur naturally in the leaf — are deliberately reduced to under 5 parts per million, because they’re allergenic and cytotoxic.
Read that last clause again. The manufacturing process exists partly to remove something the plant contains.
So this is not a leaf, and it is not close to one. It’s a pharmaceutical-grade concentrate in which the active fraction has been selected, standardised to a fixed percentage, and stripped of a native constituent nobody wants. Whatever you think about herbal medicine in general, the thing in the capsule is a manufactured product with a specification, and it should be judged like one.
Which brings us to whether it works.
Two trials, nearly six thousand people, five to six years
The question most people bring to ginkgo is whether it protects an ageing memory. That question has been tested about as thoroughly as anything in this field.
The GEM study — Ginkgo Evaluation of Memory, published in JAMA in 2008 — randomised 3,069 people aged 75 and over, with either normal cognition or mild cognitive impairment, to EGb 761 at 120 mg twice daily or placebo. They were assessed every six months for a median of 6.1 years.
It found no reduction in dementia incidence. Null.
GuidAge, published in Lancet Neurology in 2012, took 2,850 French primary care patients who had complained of memory problems and followed them for five years on the same extract at the same dose. Over that period, 61 people on ginkgo developed an Alzheimer’s diagnosis, against 73 on placebo. Not significant. The authors’ own summary: long-term use of standardised ginkgo biloba extract in this trial did not reduce the risk of progression to Alzheimer’s disease compared with placebo.
A third, smaller trial reported that ginkgo neither altered the risk of progression from normal cognition nor protected against decline in memory function.
GuidAge did produce one thing worth naming precisely: a post hoc analysis suggesting a possible effect appearing after year five. Post hoc means the analysis was chosen after the results were seen, and it is the standard way a null trial acquires a hopeful footnote. It’s a hypothesis for the next study. It is not what the trial found.
The positive evidence, and where it comes from
Here is where it gets complicated, because there is positive evidence — for treating existing mild dementia rather than preventing it — and it deserves to be described accurately rather than dismissed.
A 2014 meta-analysis pooled seven randomised placebo-controlled trials covering 2,625 outpatients over 22 to 26 weeks and concluded that EGb 761 at 240 mg daily is clinically effective, safe and well tolerated in dementia. A 2025 meta-analysis of pooled subgroup data from four trials, 782 patients with mild dementia, found 240 mg superior to placebo on cognition (p = 0.04), global assessment (p = 0.01), activities of daily living (p = 0.01) and quality of life (p = 0.02).
Now the part that belongs in the same paragraph as those results.
The 2014 meta-analysis states in its own methods that the independent statistical analysis was funded by Dr Willmar Schwabe GmbH & Co. KG — the manufacturer of the extract being evaluated — and that the statistician received the summary statistics from the study authors.
The 2025 meta-analysis in World Journal of Biological Psychiatry lists among its authors Sandra Schlaefke of the Division of Biometry, Dr. Willmar Schwabe, Karlsruhe.
A second 2025 meta-analysis, in European Psychiatry, lists M. Burkart of Global Medical Affairs, Dr. Schwabe Holding SE & Co. KG.
I want to be careful about what that does and doesn’t mean, because it’s easy to overread. Industry-affiliated research is not automatically wrong, the affiliations are openly declared in every case, and the trials being pooled were real trials. What it means is narrower and still important: the body of evidence supporting ginkgo for dementia is substantially produced by the company that sells the extract, while the two largest trials — the ones designed to answer the prevention question, funded by a US government body — came back empty.
There is also an independent assessment. In March 2025, researchers published an overview summarising every available systematic review of EGb 761 across all conditions. They found EGb 761 effective for behavioural and psychological symptoms of dementia in most reviews — and concluded that the methodological quality of those systematic reviews is low. Their conflict statement notes they worked independently of any employee of Dr. Willmar Schwabe, which tells you something about the norm in this literature.
And one fact that cuts firmly the other way, which I’d be dishonest to leave out: Schwabe supplied the extract for the GEM trial — the six-year study that found nothing. They provided the product, in blister packs with matching placebo, for the trial that failed. Whatever else is going on here, “manufacturer suppresses inconvenient research” is not an accurate description of it. This is the same pattern we found with creatine, where the researchers behind a flawed meta-analysis publicly conceded the flaw: the system is messier than a conspiracy and more compromised than a clean process.
Then there’s the question of whether you’re buying ginkgo at all
This is the part that changed how I read that pharmacy shelf.
In 2012, researchers analysed 18 ginkgo supplements bought locally in Maryland and online. Seven were clearly adulterated — spiked with rutin, quercetin, or an unidentified flavonol glycoside, with the adulterants traced to buckwheat or the Japanese pagoda tree.
The mechanism of that fraud is elegant and worth understanding, because it generalises. Quality testing for ginkgo checks flavonol glycoside content — that’s the number on the label, the “24%.” So if you take a weak or partial ginkgo extract and add cheap rutin or quercetin from an unrelated plant, the product hits its specification on paper while containing far less actual ginkgo. The test measures a proxy, and the proxy can be bought by the kilo.
Meanwhile EGb 761 has been described by independent research groups as the gold standard, on the grounds that it consistently met the specifications of past and present pharmacopoeias.
Sit with the shape of that. The extract with the best documented quality control is made by the company whose own large prevention trials failed. And a substantial fraction of the cheaper products competing with it may not contain what the label says. Neither of those facts helps the person standing at the shelf, but knowing both is better than knowing neither.
This is a sharper version of something we found with blueberry extracts: once a plant is concentrated into a capsule, the number on the label stops describing a food and starts describing a manufacturing target — and manufacturing targets can be met in ways nobody intended.
The safety signal
Ginkgo’s reputation is mild, and mostly it is. Reported side effects run to headache, palpitations, gastrointestinal upset, constipation and skin reactions.
The bleeding question is different, and the evidence is genuinely mixed in an interesting way.
In a large Veterans Administration population, people taking ginkgo concurrently with warfarin had a significantly higher risk of a bleeding adverse event: hazard ratio 1.38, 95% CI 1.20–1.58, p < 0.001. That is a real signal in a big dataset. Two caveats from the same literature: the absolute increase in bleeding events was relatively small, and a systematic review and meta-analysis found no significant effect of ginkgo on prothrombin time, activated partial thromboplastin time, or platelet aggregation.
So the clotting markers don’t move, and the outcome does. That’s an uncomfortable combination — it means we have an association without a confirmed mechanism, which is exactly the situation where honest people disagree. Alongside it sit case reports linking ginkgo use to bleeding events including severe intracranial bleeding.
A 2025 retrospective study of 2,647 prescriptions at a Vietnamese hospital found ginkgo involved in a drug interaction in 12.94% of cases, most often with clopidogrel and aspirin.
The practical guidance is unambiguous regardless of the mechanism debate: avoid ginkgo while taking any coagulation-modifying drug, and stop it at least two weeks before surgery. If you’re on warfarin, a DOAC, clopidogrel, aspirin or regular NSAIDs, this is a question for your prescriber, not a decision to make from an article. That’s the same principle as the stroke signal we covered with Alpha-GPC: a supplement with a documented interaction stops being a low-stakes experiment.
What I’d do
If your goal is preventing dementia: don’t. This isn’t a case of thin evidence or a promising signal awaiting confirmation. Two properly powered trials, nearly six thousand people, five and six years, government-funded, using the best-characterised extract in existence, both came back null. That question has been asked and answered about as well as this field ever manages.
If you have mild dementia and a doctor has suggested it: that’s a different question with different evidence, and it’s between you and them. The positive results are real, modest, in the p = 0.01 to 0.04 range, and heavily produced by the manufacturer. Worth knowing when weighing it. Not worth me overriding a clinician about.
If you’re buying it anyway: buy the standardised one. The expensive box on the shelf that specifies 24% flavone glycosides and 6% terpene lactones is describing the fraction that was actually tested in trials. The cheap box that just says “Ginkgo Biloba 120 mg” is telling you the weight of something, and roughly a third of the products in one testing round weren’t what they claimed. If you’ve decided to spend money on this, spending it on a product that matches the research specification is the only version of the decision that makes sense.
If the question was actually your memory
Everything above answers “does ginkgo work.” It doesn’t answer the question most people had before they picked up the box, which is what to do about a memory that feels less reliable than it used to.
That question does have answers, and they’re duller and better evidenced than anything on the shelf. Sleep is the one with the largest and most immediate effect — the accumulation of sleep pressure across a day, and what happens when you keep short-changing it, is the system underneath most of what we cover. Cardiovascular health is the best-evidenced long-horizon lever there is, which is unglamorous enough that nobody sells it. Physical activity has better cognitive data behind it than any compound in this category.
None of those is a purchase, which is presumably why the pharmacy shelf doesn’t stock them.
And the thing I’d take from this beyond ginkgo: I came in thinking plant extracts occupied a comfortable middle ground between food and pharmaceuticals. An acetone extraction at fifty-to-one, standardised to a fixed percentage and stripped of a compound the plant makes, is not a middle ground. It’s a drug with a botanical origin story — and the right question isn’t whether the plant is traditional, but whether the extract has been tested and what the tests found.
For this one, they have been, and they found very little. Which — and this matters more than the verdict — is what a null looks like when the trial was designed properly. Two large populations, adequate dose, five and six years, government funding, right endpoint. Most of the nulls in this field aren’t like that; most of them have an explanation that isn’t “it doesn’t work.” Ginkgo’s is.
About this article
Written by Drew Anton. Covers nootropics, stimulants, sleep and focus protocols, and wearables. Not a physician or research scientist — reads the primary literature closely and refuses to round up.
Medical review: None. NeuriFuel does not currently have a licensed clinician on the editorial team, and this article has not been medically reviewed. We state this rather than implying an authority we do not have. See our About page for our full methodology.
Sources: Built from the two large prevention trials, the manufacturer-linked meta-analyses with their funding declarations quoted, an independent overview of systematic reviews, product adulteration testing, and pharmacovigilance data on bleeding. Industry affiliation is stated next to the study it affects. Where the manufacturer’s involvement cuts in its own favour — supplying product for the trial that failed — that is stated too. One inference about a funding source that the documents do not confirm has been left out rather than implied.
Corrections: Found an error? Write to hello@neurifuel.com with a source and we will fix it and log the correction.
Last updated: 30 July 2026
References
- DeKosky ST, Williamson JD, Fitzpatrick AL, et al. (2008). Ginkgo biloba for prevention of dementia: a randomized controlled trial (GEM study). JAMA. PMC2823569 — N=3,069, median 6.1 years, 240 mg/day, null. Product supplied by Schwabe Pharmaceuticals; trial funded by NCCAM
- Vellas B, Coley N, Ousset PJ, et al. (2012). Long-term use of standardised ginkgo biloba extract for the prevention of Alzheimer’s disease (GuidAge): a randomised placebo-controlled trial. Lancet Neurol. — N=2,850, 5 years, 61 vs 73 AD cases, not significant
- Dodge HH, Zitzelberger T, Oken BS, Howieson D, Kaye J (2008). A randomized placebo-controlled trial of Ginkgo biloba for the prevention of cognitive decline. Neurology 70(19 pt 2):1809–1817 — null in unadjusted analysis
- Efficacy and tolerability of Ginkgo biloba extract EGb 761® in dementia: systematic review and meta-analysis. PMC4259871 — N=2,625, 22–26 weeks. Independent statistical analysis funded by Dr Willmar Schwabe GmbH & Co. KG
- Meta-analysis of EGb 761 in mild dementia (2025). World J Biol Psychiatry. DOI 10.1080/15622975.2024.2446830 — co-author affiliated with Division of Biometry, Dr. Willmar Schwabe
- Riepe M, Burkart M (2025). Meta-analysis of Ginkgo biloba extract EGb 761 in the treatment of mild dementia. Eur Psychiatry. DOI 10.1192/j.eurpsy.2025.1749; PMC12436692 — N=782, co-author affiliated with Global Medical Affairs, Dr. Schwabe Holding
- Pfuhlmann K, Koch AK, Langhorst J (2025). Ginkgo biloba leaf extract EGb 761® for the treatment of various diseases: overview of systematic reviews. Phytomedicine. — “Methodological quality of systematic reviews is low”
- EGb 761 specification: drug-extract ratio 35–67:1, acetone 60% solvent, 22–27% flavone glycosides, 5–7% terpene lactones, <5 ppm ginkgolic acids. PMC6397839
- Stoddard GJ, et al. Ginkgo and warfarin interaction in a large Veterans Administration population. PMID 26958257 — HR 1.38 (95% CI 1.20–1.58, p<0.001)
- Bent S, Goldberg H, Padula A, Avins AL (2005). Spontaneous bleeding associated with Ginkgo biloba: a case report and systematic review. J Gen Intern Med 20(7):657–661. PMID 16050865
- Impact of Ginkgo biloba drug interactions on bleeding risk and coagulation profiles (2025). PLOS ONE. DOI 10.1371/journal.pone.0321804; PMC11991284 — 2,647 prescriptions, 12.94% interaction prevalence
- Harnly J, et al. (2012) — 18 ginkgo supplements analysed, seven clearly adulterated with rutin, quercetin or unidentified flavonol glycoside. Reported in ABC/AHP/NCNPR Botanical Adulterants Bulletin (Gafner 2018, 2022)
- Wohlmuth H, et al. (2014) — independent assessment describing EGb 761 as consistently meeting pharmacopoeial specifications