Tinnitus Treatment: What Actually Works

There is a recurring frustration with tinnitus pages. The reader arrives having been told the sound is manageable, and leaves still hearing it. That gap is the gap between a cure and a management plan, and sellers with something to sell routinely confuse the two.

So this page is organised around one question: what has actually been shown to help? The 2025 synthesis of the clinical literature found four families of approach with consistent support. Everything else is weaker, more ambiguous, or marketed more confidently than the evidence warrants. The wider <a href="/tinnitus/">tinnitus guide</a> covers what tinnitus is and what causes it, and <a href="/tinnitus-symptoms/">the symptoms page</a> covers how to describe what you are hearing in terms a clinician finds useful. This page is about the intervention list.

This sounds defeatist until you understand what management actually aims at. Most of the approaches that work for tinnitus do not change the sound. They change three other things: how distressing it feels, how loud it seems against a quiet background, and how much of the day it takes up. Those are the variables that determine whether someone keeps working, keeps sleeping and stops withdrawing from quiet rooms.

The commercial version matters as much as the clinical one. A product that promises to remove a noise in the ear is promising something no intervention has delivered, and that is the promise to be most careful about.

An umbrella review is a review of reviews. Rather than pooling trial data, it maps the existing syntheses, looks at how consistent their conclusions are, and rates how much confidence to place in them. When 44 systematic reviews point the same way on four interventions, that is a meaningful signal rather than a single study’s finding. When they disagree, or when the underlying trials are small and heterogeneous, the conclusion is that the question is open.

The table below is the practical output of that map. Read the last column before the second one, because an approach that suits a particular person is still worth less than an approach that has been shown to work at all.

Acupuncture and dietary supplements are both absent from the strong-evidence group, and they are also the two most frequently sold. The sections on each set out what the evidence actually says rather than omitting them.

Tinnitus is unusual among medical symptoms because the patient perceives it directly and continuously, and because the perception itself becomes the problem. There is nothing to look at, nothing to measure at home, and no test result to normalise against. In that setting, the way someone responds to the sound often determines how much it costs them, and behavioural therapy works on exactly that response.

In practice it involves working out which situations are worst and why, examining the beliefs that have grown around the sound, and reversing the behaviours that make it worse. Avoiding silence because it is unbearable is the clearest example: withdrawing from quiet rooms removes the context in which the sound is loudest, which makes the sound more intrusive rather than less. CBT targets that loop directly.

It is also the approach most likely to be waved off as talking therapy. It has a defined mechanism and a measurable outcome, and asking how distress is being assessed is a fair question for any provider.

Two distinct things can be happening, and the second is often missed. A person with a high-frequency loss hears the world dulled and the tinnitus unusually sharply, because the frequencies carrying the signal have been removed while the tinnitus frequency remains exposed. Correcting the loss with amplification can restore the balance between them, and that is a genuinely different mechanism from anything that changes the tinnitus itself.

That is why the audiological assessment is not an optional preliminary. Without an audiogram, nobody can say whether hearing aids are the right tool, and a plan built on them without one is a guess. The NIDCD describes hearing aids as amplifying sound for people whose hearing loss is caused by damage in the inner ear, and the treatment literature is consistent that this is the intervention of first choice when loss is present.

The common element is a low-level sound signal delivered over weeks or months, and the common aim is to make the tinnitus less prominent rather than to drown it out completely. Complete masking has a disadvantage that people notice for themselves: the moment the sound stops, the tinnitus is unmasked and can feel more intrusive than before. Programmes that use sound at a level that leaves the tinnitus partially audible tend to work better in practice.

Masking is also the most immediately useful thing available for sleep specifically, and that is worth separating out. If the practical problem is waking at 3am to a silent room, a fan, a sound machine or a bedside generator is a reasonable first step, and it does not need to be sold as a cure. Generators fitted and monitored as part of a retraining programme are a different thing again, and better supported.

Music therapy belongs here rather than in the marketing pile. It uses structured listening rather than undifferentiated noise, which some people find tolerable where white noise or static is not, and the 2025 synthesis found consistent benefit for sound and music therapy as a category.

• Continuous low level sound. White noise, pink noise, a fan, a radio or a bedside generator. Best evidence as part of a programme rather than as a standalone long-term treatment.

• Partially masked rather than fully masked. Leaving the tinnitus slightly audible avoids the rebound effect when the external sound stops.

• Structured listening. Tinnitus retraining therapy and music therapy deliver the sound deliberately, over time, alongside counselling.

• Sleep as the first target. For most people the earliest measurable gain is better sleep, which is worth more than a change in loudness nobody can measure at home.

High heterogeneity is routinely misreported in both directions. It means the studies are measuring different outcomes, in different populations, with different methods, and reaching different answers. That is not evidence of no effect, and it is not evidence of effect. It means nobody has yet run the study that settles it.

The practical consequence is a set of questions worth asking before spending money on a course of treatment: what outcome was measured, was there a control group, how many participants were included, and how long was the follow-up. Acupuncture for tinnitus is inexpensive enough that these questions rarely get asked. Repetitive transcranial magnetic stimulation and similar device-based approaches are not, and the answers to those four questions are what separates a treatment from a purchase.

Ginkgo biloba is the herbal extract most commonly prescribed for tinnitus in Europe, which makes it the natural control case. If a heavily studied supplement in this category cannot demonstrate an effect, an unstudied blend cannot be assumed to do better, and the category’s track record at the top of it is the relevant prior.

The numbers are specific, and most summaries of this literature get them wrong.

Neither review found a significant difference in adverse effects, so this is not a story about tolerability. It is a story about an absence of demonstrated benefit, in a symptom that is known to respond to expectation. The Cochrane authors frame a strong placebo effect in tinnitus management as a feature of the evidence base, which is the reason almost every tinnitus trial carries a placebo arm and the reason uncontrolled results in this condition are worth very little.

No supplement has demonstrated an effect on tinnitus. That is the position for the whole category rather than a verdict on any individual bottle. It is also a gap rather than a demonstrated absence of effect, because supplement trials are rarely funded or published. Our <a href="/audifort-for-tinnitus/">Audifort and tinnitus page</a> sets out why absence of a trial is normal in this category, and <a href="/audifort-side-effects/">the side effects page</a> covers the interaction profiles that matter if anyone is taking a blend alongside medication.

1. Get an audiological assessment. An audiogram establishes whether there is measurable loss, which determines whether hearing aids are relevant at all. Ear examination and a review of medicines matter too, since more than 200 medicines can cause tinnitus or make it worse when started or stopped.

2. Rule out the things that are not tinnitus. Impacted earwax, middle-ear fluid or infection, otosclerosis and M&eacute;niere’s disease present differently and follow different pathways. <a href="/muffled-hearing/">Muffled hearing</a> covers the conductive side.

3. Start with the best-supported approach for the actual problem. If the sound is distressing, that is behavioural therapy. If there is measurable loss, that is amplification. If sleep is the problem, that is sound at night, delivered as part of a plan rather than as the whole plan.

4. Judge it on something other than loudness. A dated diary with one fixed daily question, scored out of ten, is a far better measure than whether the sound seems better this week, because tinnitus fluctuates naturally with sleep, stress, posture and background noise.

5. Re-measure hearing rather than relying on impression. If a change was made to treat hearing, a repeat audiogram is the only way to know whether it worked.

6. Keep the supplement question last. A supplement is not forbidden. It is a consumer choice with an unknown expected effect, and it belongs after the interventions with trial support rather than instead of them.

One group should skip all of this sequencing and be assessed promptly instead: tinnitus that keeps time with the heartbeat, tinnitus in one ear only, tinnitus with sudden hearing loss, and tinnitus with new vertigo or any neurological symptom. <a href="/pulsatile-tinnitus/">Pulsatile tinnitus</a> follows a different clinical pathway entirely, and sudden hearing loss has a two-week treatment window that no self-directed plan can act inside.

This site has an affiliate relationship with a hearing supplement, disclosed at the top of every page. This one does not end with a product recommendation, because the evidence points one way and the money points the other, and because a page that told you a bottle was the answer after three thousand words of clinical guidance would be one you were right to distrust.

Published 24 Jun 2026Last updated 25 Sept 2026