Ginkgo biloba

What it is

The special ginkgo extract used in research is called EGb 761. The EU monograph describes it precisely: drug-extract ratio 35 to 67 to 1, extraction solvent acetone 60 percent. This is not a tea but a defined preparation.

This divides the product world: approved medicines with a tested extract on one side, dietary supplements without this commitment on the other. An analysis of 63 ginkgo preparations marketed in Germany compared the information on flavone glycosides, terpene lactones and ginkgolic acids. Finding: the medicines are reliable, the dietary supplements are not.

How it is supposed to work

The common narrative runs via blood flow: ginkgo makes the blood flow more easily, the brain gets more oxygen, memory improves. The first step has been measured in humans — a meta-analysis of 18 studies with 1,985 adults found blood viscosity lowered by 1.03 millipascal-seconds.

The second step has not. No randomized study is available for an increase in cerebral blood flow in humans; the vasodilating chain of action comes from experiments on cultured endothelial cells. Between this cell and everyday life lies exactly the stretch on which the large studies found nothing.

Treatment: where the substance lies

In existing dementia lies the robust part of the effect. A meta-analysis of nine studies with 2,561 patients over 22 to 26 weeks found for EGb 761 cognition improved by -2.86 points, activities of daily living by -0.36 and the global impression with an odds ratio of 1.88 — all tied to 240 milligrams daily.

That is why ginkgo is in the German S3 dementia guideline. In version 6.0 of February 24, 2026, recommendation 69 bears the title Ginkgo biloba in the treatment of dementia, recommendation grade weak in favor, strength of consensus 97 percent. This means: may be considered.

The older Cochrane review rates the same literature more cautiously. Of 36 included studies, nine ran for six months with 2,016 patients; of the four most recently published, three found no difference from placebo and one found very large effects. In the Alzheimer’s subgroup with 925 patients, no consistent pattern. The authors’ conclusion: inconsistent and unreliable.

Prevention: this is where it stops

Preventing is something other than treating, and there the picture is reversed. The Ginkgo Evaluation of Memory Study, GEM for short, is the largest prevention study on this extract. 3,069 volunteers aged 75 and over received 120 milligrams of EGb 761 twice daily or placebo in a double-blind design, for a median of 6.1 years. 523 developed dementia, 277 under ginkgo, 246 under placebo, hazard ratio 1.12 (0.94 to 1.33).

The French GuidAge study tested the same thing in different people: 2,854 people aged 70 and over with self-reported memory problems, also 120 milligrams twice daily, over five years. 61 cases of probable Alzheimer’s dementia under ginkgo versus 73 under placebo, hazard ratio 0.84 (0.60 to 1.18). The point estimate points in the favorable direction; the interval includes no effect.

Two independent studies, two countries, two starting groups, the endpoint missed twice. A question is rarely answered more robustly. The same applies to healthy people: a meta-analysis found effect sizes of -0.04 for memory, -0.05 for executive function, -0.08 for attention.

Tinnitus, dizziness and walking distance

For tinnitus, the picture is clear and unfavorable. The 2022 Cochrane analysis included 12 studies with 1,915 participants. On the Tinnitus Handicap Inventory, after three to six months there was a mean difference of -1.35 on a scale from 0 to 100 (-8.26 to 5.55), based on two studies with 85 participants.

For dizziness, the decisive arm is missing: the most-cited study put 160 patients on EGb 761 or betahistine for 12 weeks, without placebo. Both groups improved equally — this way, an effect cannot be separated from the spontaneous course.

In peripheral artery disease, the effect narrowly misses significance: 14 studies with 739 participants, for absolute walking distance 11 studies with 477 participants, converted 64.5 meters (-1.8 to 130.7), p = 0.06.

What the studies show

The meta-analysis on treatment

Nine randomized studies with EGb 761 in cognitive impairment and dementia, 22 to 26 weeks, 2,561 patients. The change in cognition favored EGb 761 with a weighted mean difference of -2.86 (-3.18 to -2.54), activities of daily living with -0.36 (-0.44 to -0.28), the clinical global impression with an odds ratio of 1.88. All effects were tied to 240 milligrams daily, stronger for neuropsychiatric symptoms, not present in the pure Alzheimer’s subgroup. What was measured were test scores over half a year, not the course over years.

GEM — the largest prevention study

Randomized, double-blind, placebo-controlled, at five US centers from 2000 to 2008. 3,069 volunteers aged 75 and over, 2,587 of them with normal cognition and 482 with mild cognitive impairment; 1,545 received 120 milligrams of EGb 761 twice daily, 1,524 placebo, examination every six months, diagnosis by an expert panel. Median follow-up 6.1 years, dropout rate 6.3 percent, primary endpoint dementia incidence. Result: 523 cases, 277 under ginkgo and 246 under placebo, 3.3 versus 2.9 per 100 person-years, hazard ratio 1.12 (0.94 to 1.33). Endpoint not reached, cognitive trajectories not different in any of the five domains.

GuidAge — the second independent test

Randomized, double-blind, placebo-controlled, in France. 2,854 adults aged 70 and over who had spontaneously reported memory problems to their family doctor; 1,406 received at least one dose of EGb 761 (120 milligrams twice daily), 1,414 placebo, five years of follow-up, primary endpoint conversion to probable Alzheimer’s dementia. Result: 61 cases under ginkgo (1.2 per 100 person-years) versus 73 under placebo (1.4), hazard ratio 0.84 (0.60 to 1.18); the risk was not proportional over time. Adverse events evenly distributed, 76 versus 82 deaths. Endpoint not reached. Funded by the manufacturer Ipsen.

Where the data stop

The most widely read sentence about ginkgo — that it increases blood flow in the brain — has not been demonstrated in humans. The contradiction between the data sets remains open: studies with hard endpoints over years find nothing, studies with scale scores over 22 to 26 weeks find something. An advantage on a test scale says nothing about six years of progression — and a missing preventive effect does not rule out a symptomatic one. The basis is thin precisely where the most is promised: in vascular dementia three studies with 283 patients with extreme heterogeneity, in tinnitus 85 participants in the main comparison, in dizziness no placebo arm at all. And almost everything applies to EGb 761, not to ginkgo in general.

Status, approval and legal

In Germany, dry ginkgo leaf extract is an approved medicine. The EU monograph lists it under well-established use for the improvement of age-related cognitive impairment and of quality of life in mild dementia: single dose 120 to 240 milligrams, daily dose 240 milligrams, at least 8 weeks. The leaf powder is listed separately as a traditional use for heaviness in the legs and cold hands and feet. The distinction from dietary supplements has been settled in court: on August 24, 2022, the German Federal Administrative Court ruled (file number 3 B 36.21) that products with monograph-compliant dry extract and 100 milligrams per day are medicinal products by function. Health claims for botanicals have not been conclusively assessed in the EU; ginkgo is not on the WADA prohibited list.

Safety

Tolerability has been tested over years and in thousands of participants: in GEM and GuidAge, adverse events occurred equally often in both groups, and Cochrane also found no differences from placebo. On bleeding risk there is a dedicated meta-analysis of 18 studies with 1,985 adults: no significant effect on platelet aggregation, fibrinogen, activated partial thromboplastin time or prothrombin time; for warfarin, too, a review of nine studies found no change in the anticoagulant effect. Running the other way is a finding from GEM: of 24 hemorrhagic strokes, 16 occurred under ginkgo and 8 under placebo, not significant, but worth mentioning. The European monograph maintains the warning: as a precaution, a gap of 3 to 4 days before surgery, monitoring with anticoagulants. In epilepsy, it cannot be ruled out that ginkgo promotes further seizures; contraindications are hypersensitivity and pregnancy. In two-year studies of the National Toxicology Program, ginkgo leaf extract caused liver tumors in mice; in 2016, IARC classified it as possibly carcinogenic (Group 2B), at animal doses far above those used here. Ginkgo seeds are something else: they contain a vitamin B6 antagonist, and seizures have been documented after larger amounts.

Sources

• DeKosky et al., JAMA 2008 (GEM study)
• Vellas et al., Lancet Neurology 2012 (GuidAge)
• Snitz et al., JAMA 2009 (GEM, cognitive trajectory)
• Birks & Grimley Evans, Cochrane Database of Systematic Reviews 2009
• Tan et al., Journal of Alzheimers Disease 2015 (EGb 761 meta-analysis)
• Laws et al., Human Psychopharmacology 2012 (healthy people)
• Sereda et al., Cochrane Database of Systematic Reviews 2022 (tinnitus)
• Kellermann & Kloft, Pharmacotherapy 2011 (bleeding risk)
• EMA/HMPC, European Union herbal monograph on Ginkgo biloba L., folium
• BVL on the decision of the German Federal Administrative Court of August 24, 2022 (German)