Ginkgo biloba does not appear to help tinnitus: two Cochrane reviews and a meta-analysis of six placebo-controlled trials found no reliable benefit over placebo, although one 2026 national guideline says it may be considered. It is probably the single most recommended over-the-counter supplement for tinnitus, sold for that purpose in pharmacies and health food stores worldwide and inside branded tinnitus formulas such as Arches, and the evidence behind it does not support the reputation.
What the Cochrane reviews found
Cochrane has reviewed ginkgo for tinnitus twice, and neither review found that it helps.
The 2013 Cochrane review included four randomised trials with 1,543 participants, and assessed every one of them as being at low risk of bias. Three of the trials, with 1,143 participants, enrolled people whose main complaint was tinnitus, and in them there was no evidence that ginkgo biloba was effective. The fourth enrolled 400 people with mild to moderate dementia, some of whom had tinnitus. Their tinnitus ratings started low, at 1.7 to 2.5 on a 10-point scale, and fell by a statistically significant 1.5 points in the vascular dementia group and 0.7 in the Alzheimer's group — a change whose practical significance the reviewers called unclear. Side effects were infrequent. The conclusion: the limited evidence does not demonstrate that ginkgo biloba is effective for tinnitus when tinnitus is the primary complaint.
The 2022 Cochrane review is Cochrane's current word on the question. It cast a wider net — 12 randomised trials with 1,915 participants, 11 of them comparing ginkgo with placebo — and landed on "very uncertain" rather than "no". Risk of bias across these trials was mostly high or unclear, and outcomes were measured too differently to pool much. The one pooled result on tinnitus severity rested on two studies and 85 people: a difference of −1.35 points on the 100-point Tinnitus Handicap Inventory (95% CI −8.26 to 5.55), consistent with little or no effect, at very low certainty. Bleeding, seizures and other adverse effects were no more common than on placebo, at low certainty. The authors conclude that the benefits and harms are uncertain, and ask future trials for the highest-quality randomisation and blinding because tinnitus is so prone to a placebo response.
Read together, the two reviews say the same thing from different ends. The trials at low risk of bias found no benefit for people whose main complaint was tinnitus; the wider trial base is too weak to show one.
A trial and meta-analysis that said no
A 2004 study both ran its own randomized, placebo-controlled trial and combined it with five other similar trials into a meta-analysis covering more than 1,000 patients total. The result: no statistically significant difference between ginkgo biloba and placebo on any of the outcome measures tested — tinnitus handicap scores, general health status, or hearing thresholds. In the pooled data, 21.6% of ginkgo-treated patients reported improvement, compared with 18.4% on placebo (odds ratio 1.24, 95% CI 0.89 to 1.71) — a difference small enough to be explained by chance. The authors' summary is direct: ginkgo biloba does not benefit patients with tinnitus.
American Family Physician's evidence summary of that paper put the bottom line plainly: no evidence from six randomised trials of more than 1,000 patients that ginkgo biloba is an effective therapy for tinnitus.
The most recent placebo-controlled trial points the same way. A 2025 double-blind trial in The International Tinnitus Journal randomised 46 adults with tinnitus of four weeks to a year to the standardised extract EGb 761 (240 mg a day) or a matching placebo for 12 weeks. Tinnitus Handicap Inventory scores fell in both groups — from 34.2 to 25.0 on ginkgo and from 34.4 to 22.2 on placebo — with no significant difference between them (p = 0.51). It is a small trial, and its value is the comparison most of the positive literature lacks: the placebo group improved as much as the ginkgo group.
A 2025 network meta-analysis ranked it first, and that needs explaining
If you search this question now you will find a 2025 systematic review and network meta-analysis in BMJ Open — 60 randomised trials from 21 countries — in which Ginkgo biloba with vitamin ranked first for tinnitus severity, with a standardised mean difference of −3.11 against placebo. Headlines have been written from that sentence. It is the same analysis the drug trials article reads for every other compound it ranked.
Read the rest of the paper and it says something quite different.
Its own certainty rating is "very low". The authors assessed the severity network with GRADE and that is the grade they gave it — the bottom of the scale, meaning the true effect is likely to be substantially different from the estimate. They also report three significant inconsistent comparisons inside that same network, which is the statistical sign that the trials being pooled disagree with each other in ways the model cannot reconcile.
The trials are not good. Of the 60 included, only 22% were at low risk of bias; 58% raised some concerns and 20% were at high risk. Heterogeneity for the severity network was I² = 0.67. Egger's test found significant publication bias in the loudness network — the signature of small negative trials never being published.
And the effect size is the tell. An SMD of −3.11 would be one of the largest treatment effects in medicine. For scale, on this site: the tDCS meta-analysis reports −0.35 for loudness and −0.50 for distress, and describes those as small to moderate. A supplement sold in supermarkets producing an effect six times larger than transcranial stimulation, in trials that are mostly not blinded well, is not a discovery. When an effect size is far outside the range everything else in a field produces, the usual explanation is the trials rather than the treatment.
The authors' own conclusion is appropriately hedged — that these "could be promising" and that "further trials with rigorous design and larger sample sizes are necessary". That is a call for better evidence, not a finding that ginkgo works.
None of this makes the 2025 paper worthless. A network meta-analysis is a good way to map what a field has tested, and this one maps a field whose trial base is thin, inconsistent and probably selectively published. That is a real result. It is just not the one the ranking implies.
Why "ginkgo" on a label doesn't mean one consistent thing
Most of the meaningful trial data on ginkgo for tinnitus specifically tested a single, tightly standardized extract called EGb 761 — manufactured with a defined extraction process and adjusted to a consistent 22–27% flavone glycosides and 5–7% terpene lactones (the compounds, including bilobalide and several ginkgolides, thought to be behind ginkgo's pharmacological effects). This matters because not every ginkgo supplement on a shelf uses this same standardized formulation or concentration — generic, non-standardized ginkgo products may contain meaningfully different amounts of the active compounds, which makes it hard to generalize confidently from EGb 761 trial results to whatever a specific bottle contains. If the negative trial results are the reason you're inclined to skip ginkgo, that reasoning applies most directly to EGb 761 and formulations standardized to match it — an unstandardized product is, if anything, even less likely to have any effect, not more.
The proposed mechanism, for context, centers on improved blood flow: ginkgo compounds are thought to have vasoregulatory effects (widening blood vessels and reducing platelet clumping) that could theoretically improve circulation to the cochlea, which is sensitive to blood flow given its high metabolic demands. This is a biologically plausible mechanism — it's part of why ginkgo was tried for tinnitus in the first place — but a plausible mechanism doesn't override what the clinical trials found once it was tested directly in humans.
A note on newer research: studies continue to investigate EGb 761 for tinnitus, including work looking at whether particular groups of patients respond better than the pooled average suggests. A 2025 study of 170 people taking EGb 761 (120 mg twice a day for 24 weeks) is the clearest recent example. It reported significant improvement on every tinnitus measure compared with baseline, and more improvement in people with high anxiety or stress or with normal hearing. It was exploratory, open-label and uncontrolled, so it cannot separate the extract from time and expectation; the overall response rate on its own responder criterion was 18.8%; and its disclosures state that three of its authors received grants and honoraria from the extract's manufacturer and a fourth is its employee. Subgroup findings from an uncontrolled study are hypotheses to test, not evidence that a subgroup benefits.
A 2026 study in Acta Oto-Laryngologica is a useful illustration of why newer work keeps failing to settle this. It followed 195 patients given EGb 761 after sudden sensorineural hearing loss, and the numbers look spectacular: mean Tinnitus Handicap Inventory scores fell from 48.99 to 28.33 at one month, 16.97 at six months and 10.54 at eighteen. Over 90% met the improvement criterion in both study arms.
And it cannot tell you whether ginkgo contributed anything at all, because every participant received it. The trial compared 240 mg/day against 120 mg/day — two doses of the same extract, with no placebo arm and no untreated arm. The between-group differences were not statistically significant for THI reduction or any other outcome, so the two doses performed alike.
Tinnitus after sudden hearing loss improves substantially on its own over a year. A study with no comparison group cannot separate that from a drug effect, and a near-universal "improvement rate" in a condition that mostly improves anyway is not a finding about the drug. This is the shape of most of the positive ginkgo literature: real patients, real improvement, and no way to attribute it.
Why the reputation persists anyway
A few things keep ginkgo's reputation alive despite the negative trial data:
- Older, smaller, lower-quality studies occasionally found positive results, and get cited more than the larger negative ones that came later.
- Tinnitus fluctuates naturally, and supplements are often started during a particularly bad stretch — improvement that would have happened anyway gets attributed to whatever was being taken at the time.
- Ginkgo has documented benefits in unrelated areas (mild cognitive symptoms, circulation-related conditions), and that general reputation as "good for blood flow to the brain" gets applied to tinnitus by extension, even though the specific tinnitus trials don't back it up.
- It's a low-risk, easy-to-try option, which makes it an easy default recommendation even without strong evidence — a fair thing to know before spending money on it, not necessarily a reason it's harmful to try.
Is it dangerous to try anyway?
Ginkgo biloba is generally well-tolerated — in the placebo-controlled tinnitus trials the 2022 Cochrane review examined, bleeding, seizures and other adverse effects were no more common than on placebo, at low certainty. The standard caution is about combinations: it can interact with blood thinners (including aspirin and warfarin) and increase bleeding risk, and isn't recommended before surgery. If you're not on blood thinners and want to try it despite the evidence, the risk is mainly financial rather than medical — but going in with accurate expectations, rather than the marketing claim, is the more useful starting point.
Where the evidence is better spent
If the underlying interest is trying something with real supporting research, cognitive behavioral therapy and sound therapy both have meaningfully stronger evidence for reducing tinnitus-related distress than ginkgo has for reducing tinnitus itself.
Where the professional guidelines land
This is not a lone reading of the trials. The American Academy of Otolaryngology–Head and Neck Surgery's clinical practice guideline on tinnitus issues an explicit recommendation against ginkgo biloba, melatonin, zinc and other dietary supplements for persistent, bothersome tinnitus. That is the same conclusion the pooled trial data supports, arrived at independently by a guideline panel. The other supplements it names have their own articles: melatonin, and magnesium and zinc.
The AAO-HNS guideline dates from 2014, so it predates some of the newer work discussed above. The US Department of Veterans Affairs and Department of Defense guideline of 2024 does not: it suggests against the use of ginkgo biloba, dietary or herbal supplements, or nutraceuticals for tinnitus management, as a weak recommendation against. How the two documents compare is covered in the VA/DoD tinnitus guideline. Nothing published since has overturned the headline result.
A 2026 guideline disagrees, and the disagreement is traceable
In 2026 a Korean national clinical practice guideline for tinnitus was published in Clinical and Experimental Otorhinolaryngology — a multidisciplinary panel of eight otorhinolaryngologists, a neurologist, a psychiatrist and an audiologist, working from GRADE, nine clinical questions in PICO format and a modified Delphi process requiring 80% agreement.
On supplements it splits in two directions. Routine zinc and vitamin supplementation is strongly recommended against. But Ginkgo biloba "may be considered" for patients with subjective tinnitus, at Grade B.
That directly contradicts the recommendations above, and this article is not going to bury it because it sits awkwardly with the position taken above.
What can be said about the disagreement, without pretending to settle it:
The timing points at one paper. The Korean panel searched the literature up to 2025, so it had the network meta-analysis dissected earlier on this page — the one that ranked Ginkgo biloba with vitamin first for tinnitus severity. The AAO-HNS guideline was written in 2014 and did not. Two guidelines reaching different conclusions eleven years apart are not disagreeing about the same evidence.
The Korean guideline rates that evidence as weak, in its own words. Its stated conclusion includes that most available evidence was predominantly of low or very low certainty. And "may be considered" is the weakest actionable form a recommendation comes in — it is not a recommendation to use something.
Nothing new was tested. No trial published since has overturned the pooled result at the top of this page. What changed is which review a panel had in front of it, not what the trials found.
So the position now is: one national guideline says ginkgo may be considered, two advise against it, the newest rests substantially on an analysis whose own GRADE rating is "very low" and in which Egger's test found publication bias, and the underlying trial data is unchanged. That is a live disagreement inside a weak evidence base — not a vindication.
None of which changes the practical advice here, because the practical advice never rested on the guideline. Ginkgo is inexpensive and, if you are not taking blood thinners, low risk. The point has always been going in with the expectation the trials support rather than the one the label implies.