For chronic tinnitus without sudden hearing loss or Ménière's disease, steroid injections through the eardrum have not beaten placebo: four placebo-controlled trials pooled in a meta-analysis found no improvement, though uncontrolled studies report far better numbers.
This treatment is worth covering in detail because it shows something important about reading medical research generally: the gap between what uncontrolled studies report and what properly controlled trials find can be enormous.
What the injection is
Intratympanic steroid injection delivers a corticosteroid (most commonly dexamethasone) directly through the eardrum into the middle ear, where it can diffuse into the inner ear at higher local concentrations than oral steroids typically achieve. This same technique is well-established and effective for certain conditions — it's a standard part of treatment for sudden sensorineural hearing loss and is used for Ménière's disease. Its use for chronic subjective tinnitus, without a sudden hearing loss or Ménière's diagnosis, is a separate, more specific question.
The striking gap between study types
Uncontrolled studies — meaning studies that gave patients the injection and simply measured before-and-after results, without a placebo comparison group — have reported dramatic results. One frequently cited Japanese series, covering more than 3,000 patients, reported tinnitus improving in 75% of patients immediately after treatment and in 68% at six months. The same authors found the effect varied by cause: good in labyrinthine hydrops, the inner-ear fluid build-up associated with Ménière's disease, and poor for tinnitus caused by noise or drugs. Numbers like 75% are exactly what makes a treatment sound compelling in isolation.
A 2022 systematic review and meta-analysis pooled the double-blind, placebo-controlled randomized trials — the four studies designed to isolate the injection's real effect from placebo response, natural fluctuation, and regression to the mean — and found something very different: no evidence of tinnitus improvement compared with placebo (odds ratio 1.38, 95% confidence interval 0.53 to 3.61, wide enough to include no effect at all).
The same review also gathered the less rigorous studies. Seven retrospective studies without controls reported improvement rates ranging from 35.9% to 91.3%. Four prospective studies with controls suggested the injection might help when combined with other drugs, which the authors flagged for further research. That is the only opening the review leaves: steroid injection on its own did not beat placebo.
A trial that tested it against saline
A randomised, single-blind trial published in 2005 took 36 patients with severe, disabling tinnitus, mostly of cochlear origin, and gave them four weekly injections of either dexamethasone or plain saline. Immediately after treatment, 33% of ears in the dexamethasone group and 29% in the saline group had improved by at least 2 points on a visual analogue scale — not a significant difference. Over follow-up of 13 to 31 months, those who had improved drifted back to where they started.
The authors' conclusion is the useful one: there was no advantage for dexamethasone over saline, and both produced a placebo-like improvement. Injecting salt water through the eardrum helped almost as many people, briefly, as injecting a steroid.
Why this gap happens
Tinnitus fluctuates naturally, responds to attention and expectation, and tends to regress toward a more typical level over time regardless of what's done to it. So almost any intervention tested without a placebo comparison will show some patients "improving" through these unrelated effects. The saline arm above is that effect made visible: nearly a third of ears improved on an inactive injection. When a controlled trial and an uncontrolled series disagree this sharply, the difference between them is a measure of how much of the uncontrolled figure was never about the drug.
What this means practically
If intratympanic steroid injection is suggested to you for chronic tinnitus without a sudden hearing loss or Ménière's diagnosis, it's worth asking what evidence your provider is drawing on — the uncontrolled study numbers, or the controlled-trial evidence. This isn't a treatment with zero rationale (the underlying idea of reducing inner ear inflammation is coherent), but the properly controlled evidence hasn't supported the impressive-sounding uncontrolled results, and that matters before a procedure that, unlike sound therapy or CBT, involves an injection through the eardrum.
The same question works on any tinnitus treatment claim: a dramatic percentage from an uncontrolled study and a modest or null result from a controlled trial are not equally reliable evidence, even though marketing often presents only the more impressive number. Reading a tinnitus study sets out the patterns to look for.
Where the professional guidelines land
The American Academy of Otolaryngology–Head and Neck Surgery's clinical practice guideline on tinnitus issues a recommendation against intratympanic medications for the routine treatment of persistent, bothersome tinnitus — alongside recommendations against antidepressants, anticonvulsants and anxiolytics for the same purpose.
The 2024 VA/DoD guideline reaches the same place. It suggests against intratympanic corticosteroid injections for tinnitus management (weak against), in the same recommendation as anticonvulsants, antidepressants, antiemetics, antithrombotics, betahistine and NMDA receptor antagonists.
Both are specifically about tinnitus as the treatment target. Intratympanic steroids have a separate and much better-supported role in sudden sensorineural hearing loss, where timing matters. Same injection, different question.
The other injection offered for tinnitus
If you are weighing an injection at all, there is a second one in circulation that works on a completely different target: a stellate ganglion block, which puts local anaesthetic near the sympathetic nerves in the neck rather than into the ear. It has its own 2026 meta-analysis, its own encouraging headline numbers, and its own reasons for caution — chiefly that the pooled improvement in tinnitus handicap lands below the threshold at which patients notice a change.