There is no pill for tinnitus: no medication is approved anywhere for treating tinnitus itself — not in the US, not in Europe — and the drugs tried against it have not held up in controlled trials. Every drug taken in connection with tinnitus is being used for something adjacent to it, or off-label on evidence that did not hold.
This article covers what was tried and what happened, because "there's no pill" is easier to accept when you can see the trials it came from.
One clarification first, because the vocabulary collides: this is about drugs tried against tinnitus. Drugs that cause or worsen it are a different list entirely, covered on the medications linked to tinnitus page, in the article on ototoxic medications, and, for the erectile dysfunction drugs whose labels warn of sudden hearing loss, in Viagra, Cialis and sudden hearing loss.
What the guidelines say
The American Academy of Otolaryngology–Head and Neck Surgery's clinical practice guideline on tinnitus makes a recommendation against antidepressants, anticonvulsants, anxiolytics and intratympanic medications for the routine treatment of persistent, bothersome tinnitus.
Four drug classes, one recommendation, all pointing the same way. That is the shortest summary of decades of pharmacological effort on this symptom.
The 2024 guideline from the US Department of Veterans Affairs and Department of Defense reaches the same place with a longer list and a weaker grade. It suggests against anticonvulsants, antidepressants, antiemetics, antithrombotics, betahistine, intratympanic corticosteroid injections and NMDA receptor antagonists for tinnitus management, in a single weak recommendation against. The VA/DoD guideline and betahistine each have their own article.
Anticonvulsants: the class with the most trials behind it
Anticonvulsants were a reasonable idea. If tinnitus involves hyperactive neural firing, a class of drugs that stabilises overactive neurons is exactly what you would try.
A Cochrane review pooled seven trials covering 453 patients, across gabapentin, carbamazepine, lamotrigine and flunarizine. The findings, in order of what matters:
- Three studies used a validated questionnaire as their primary outcome. None showed a significant positive effect.
- One of those studies found a significant negative effect: gabapentin raised Tinnitus Questionnaire scores by 18.4 points compared with placebo — the tinnitus got worse.
- Pooling the gabapentin data across studies produced no effect (SMD 0.07, 95% CI −0.26 to 0.40).
- A secondary analysis of "any positive effect" on a self-assessment score did find a small favourable difference (14%, 95% CI 6% to 22%). An analysis of "near or total eradication of tinnitus annoyance" found nothing (4%, 95% CI −2% to 11%).
- 18% of patients experienced side effects.
The reviewers' own conclusion is that the existing evidence carries significant risk of bias, that there is no evidence of a large positive effect, and that the small effect demonstrated is "of doubtful clinical significance".
Worth holding onto that split: the only positive signal came from asking people whether anything felt better, not from the validated instruments, and it disappeared entirely once the question became whether the tinnitus had substantially gone.
Lidocaine: the finding that goes nowhere
Intravenous lidocaine is the most interesting negative result in tinnitus pharmacology, because it is not actually negative — it just cannot be used.
In a study of 30 patients given lidocaine by intravenous infusion at 1.5 mg/kg over 30 minutes, four patients (13.3%) reported total suppression of their tinnitus immediately afterwards, and three more (10%) reported partial relief. The researchers also measured otoacoustic emissions before and after, finding statistically significant changes in cochlear responses.
So roughly a quarter of the group had their tinnitus measurably interrupted by a drug. And it is not a treatment, for reasons that are not going to be solved:
- It requires intravenous infusion, not a tablet.
- The suppression is temporary.
- Lidocaine is an antiarrhythmic with real cardiac effects, which makes repeated infusion for a non-life-threatening symptom impossible to justify.
Its value is the same as the value of residual inhibition: proof that the phantom signal can be switched off, and therefore that it is a signal rather than a permanent structural fact. Nobody has converted either finding into something you can take home.
It is also worth keeping in mind when reading about any procedure that delivers a local anaesthetic near the head and neck. A stellate ganglion block is offered for tinnitus on the theory that interrupting sympathetic nerve signalling improves blood flow to the ear — but since the procedure injects anaesthetic into a well-vascularised part of the neck, any short-term suppression it produces has a second possible explanation sitting right here in this section.
Where a drug does have a role
None of the following is treating the tinnitus signal. All of them are worth knowing, because "no drug treats tinnitus" gets misread as "there is no point discussing medication with a doctor".
For co-occurring depression or anxiety. SSRIs are prescribed for the mood and anxiety that frequently accompany chronic tinnitus, and improving those can substantially change how much the tinnitus intrudes without altering the sound at all. That article also covers the other direction — tinnitus is listed as a possible side effect of nearly every SSRI.
For one specific syndrome. Typewriter tinnitus — a rare, staccato, one-sided presentation caused by a vessel compressing the auditory nerve — responds to carbamazepine in most reported patients. This is not an exception to the guideline; the guideline's subject is routine tinnitus, and this is a defined syndrome with an identified mechanism.
For an underlying condition. If the tinnitus is downstream of something treatable — an infection, a thyroid problem, anemia, high blood pressure — treating that is the intervention, and any tinnitus improvement follows from it.
Not for sleep, indefinitely. Benzodiazepines are sometimes prescribed for the sleeplessness that comes with new tinnitus. They work, and they carry dependence risk, and stopping them after long-term use has its own documented association with tinnitus during withdrawal. That is a genuinely difficult trade-off rather than a simple no, and the evidence on tinnitus and insomnia shows how little of the case for sleeping pills comes from tinnitus research.
What 60 trials look like when you pool them all
A 2025 systematic review and network meta-analysis in BMJ Open did the most complete thing anyone has done with this literature: it pooled 60 randomised controlled trials from 21 countries, comparing pharmacotherapies against each other and against placebo, with tinnitus severity as the primary outcome.
The rankings it produced are widely quoted. What sits underneath them is more useful.
- Only 22% of the 60 trials were at low risk of bias. 58% raised some concerns, 20% were at high risk.
- The severity network was graded very low certainty on GRADE, with three significant inconsistent comparisons inside it.
- Heterogeneity for severity was I² = 0.67 — the included trials substantially disagree.
- Egger's test found significant publication bias in the loudness network.
The compounds that ranked first — Ginkgo biloba with vitamin for severity, acamprosate for annoyance, fluoxetine for loudness — did so with effect sizes that should themselves raise an eyebrow. Ginkgo's standardised mean difference of −3.11 and fluoxetine's of −3.28 are several times larger than anything else in this field achieves; for comparison, tDCS reports −0.35 and −0.50 and those are described as small to moderate.
The authors' conclusion is that these "could be promising" and that further trials with rigorous design and larger samples are necessary. Read precisely, this is a map of a weak literature rather than a shortlist of things that work — and it is the strongest overview of tinnitus pharmacotherapy currently available.
Two drug combinations, and a lesson in how to read a trial
A 2025 randomised placebo-controlled trial in Otolaryngology–Head and Neck Surgery tested two drug pairs against placebo: nortriptyline with topiramate, and verapamil with paroxetine.
Start with the within-group results, because these are the ones that get quoted:
- Nortriptyline–topiramate (n = 19): Tinnitus Functional Index fell from 58.4 to 46.3 — p < .001
- Verapamil–paroxetine (n = 22): TFI fell from 54.6 to 42.2 — p = .004
- Placebo (n = 26): TFI fell from 51.2 to 45.2 — p = .086, not significant
Two active arms improve significantly. Placebo does not. That reads like a finding.
Now the comparison that decides it. Between the arms there was no statistically significant difference at all — ANOVA p = .265.
Look again at what placebo did: it dropped six points. The reason that change missed significance is not that nothing happened to the placebo group; it is that the drop was somewhat smaller and the group's spread wider. A treatment reaching significance against its own baseline while placebo does not is not the same as the treatment beating placebo, and only the second question tells you whether a drug works.
This is the single most common way a tinnitus result gets misreported, and it is worth being able to spot, because the same structure turns up constantly in supplement and device marketing: two within-group p-values presented side by side, and no between-group comparison anywhere. The authors here are careful — their own conclusion is that both combinations were promising and that larger trials with longer follow-up are needed to validate the findings between groups.
Memantine, and the same pattern again
Memantine — an NMDA-receptor blocker used in Alzheimer's disease — comes up because the receptor it acts on is implicated in the runaway excitability theories of tinnitus. A 2025 trial in Brain and Behavior tested it.
The reported result: after treatment, tinnitus scores on both a numerical rating scale and the Tinnitus Severity Index were significantly lower in the memantine group, and the authors conclude that memantine "may have remarkable effect in reducing tinnitus and its discomfort".
Read one sentence of their own results before accepting that. The numerical rating scale improved significantly in both groups — memantine and control alike — and only the Tinnitus Severity Index separated them. So one of the two outcome measures showed the control group improving just as much.
That is the same structure as the combination trial above: a confident conclusion resting on the outcome that separated, with the outcome that did not left in the results and out of the summary. When a paper reports two measures of the same thing and only quotes one, the other is the one to look for.
Memantine is not currently supported as a tinnitus treatment by anything stronger than this, and it belongs to the class — NMDA receptor antagonists — that the VA/DoD guideline suggests against.
Why nothing general has worked
The most likely reason is that "tinnitus" is not one condition. A phantom sound following blast exposure, one accompanying Ménière's disease, one driven by a jaw problem and one arriving with age-related hearing loss share a symptom and not a mechanism. A drug developed against any single mechanism is being tested, in most trials, on a population where most participants have a different one.
That is also why the current wave of drug development is narrower by design — targeting cochlear synaptopathy specifically, or shared neural hypersensitivity in people who also have migraine, rather than "tinnitus". A placebo-controlled phase 1b/2a trial of CIL001 (paliroden), a single injection through the eardrum in about 60 adults with constant tinnitus in both ears and difficulty hearing speech in noise, has a planned start in January 2027. COMPACT-PM, at Stanford, plans to compare anti-CGRP migraine drugs with conventional migraine preventives in about 120 adults with bothersome tinnitus and a history of migraine, with no placebo arm. Both were registered in 2026, and when their ClinicalTrials.gov records were checked in September 2026, both were listed as not yet recruiting. Gene therapy, which has restored hearing in children with a rare inherited deafness, addresses a different problem from tinnitus. Where tinnitus research is headed follows these candidates: none is available, and most drugs at this stage never reach the market.
What to do with this
- Do not go looking for a prescription for tinnitus itself. There is nothing to ask for, and the one class with the most trials behind it produced a trial where the tinnitus got measurably worse.
- Do raise mood, anxiety and sleep with a doctor, which are treatable, are frequently the part doing the real damage, and are where medication has a legitimate role.
- Do describe your tinnitus precisely. The narrow exceptions above are identified by pattern — staccato and one-sided, or episodic with vertigo and fullness — not by asking whether anything can be done.
- Do not wait for the drug you read about. The compounds developed specifically for tinnitus — AM-101, OTO-313 and the rest — have reported their controlled trials, and every one of them failed its primary endpoint. A press release outlives a result.
- Treat any product marketed as a tinnitus drug or supplement with the guideline in mind: ginkgo, melatonin and zinc are named in the same document, in the same direction. The supplements articles cover why.