Whether acupuncture helps tinnitus is still unresolved: trials comparing it with sham (fake) acupuncture have mostly failed to show a benefit, while trials against waiting lists or other treatments look positive but cannot rule out placebo effects. It is one of the most commonly tried complementary approaches for tinnitus, and the reason it is hard to give a clean verdict on is not that the results are clearly negative, as they are for ginkgo biloba, but that trial quality across the field has been consistently weak.
What the systematic reviews found
Several independent systematic reviews have looked at this question, and the pattern across them is a genuine "unclear" rather than a clear yes or no:
- An early systematic review of 9 randomized trials found that in the five comparing acupuncture or electroacupuncture against sham (fake) acupuncture, the results failed to show a statistically significant improvement — an important comparison, since it's designed to control for placebo and the general experience of receiving hands-on care. The exception was two small trials of a single, brief scalp acupuncture session, which beat a sham needled at non-acupuncture points on a scale of symptom relief. The authors described the trials' methodological quality as mostly poor, and concluded they were too few, too small and too weak for definitive conclusions.
- A more recent meta-analysis found acupuncture had no significant effect on one primary tinnitus loudness measure (a visual analogue scale) compared to control treatment, but did find positive effects on two other standard tinnitus questionnaires (the Tinnitus Handicap Inventory and Tinnitus Severity Index) — a split result across different outcome measures within the same analysis, which is itself a sign of how inconsistent this evidence base is.
- That same review explicitly concluded that due to the low quality and small sample size of the included trials, the evidence wasn't sufficient to draw definitive conclusions either way, and called for larger, more rigorous studies.
- A review of electroacupuncture specifically (acupuncture combined with mild electrical stimulation of the needles) found five small studies with 322 participants, judged all five at high risk of selection and performance bias, and could not pool them; it found no convincing evidence that electroacupuncture helps.
Scalp acupuncture, and why its strong numbers do not settle it
A 2025 systematic review and meta-analysis in Complementary Therapies in Medicine examined one specific variant — scalp acupuncture, needling points on the head rather than the body — across 20 studies and 1,430 participants. Its results are the most positive in this article:
- Clinical effective rate — RR 1.25 (95% CI 1.16–1.35), with low heterogeneity (I² = 20%)
- Tinnitus severity — SMD −0.76 (−1.02 to −0.51), heterogeneity I² = 62%
- Scalp acupuncture plus auxiliary acupoints — SMD −0.93 (−1.52 to −0.33), heterogeneity I² = 86%
An SMD of −0.76 is a large effect by this field's standards — roughly double what tDCS achieves for distress. So why does it not change the verdict above?
Because of what it was compared against. The review states its comparison plainly: scalp acupuncture against traditional acupuncture or pharmacotherapy. Those are active comparators, not sham. A result showing that needling the scalp outperforms needling the body tells you which of two acupuncture styles scored higher in unblinded trials. It cannot tell you whether either one beats the experience of being treated — and that comparison, acupuncture against sham, is the one that has mostly failed to show an effect.
The authors do not overstate it. They grade the certainty of the evidence as moderate to low, report substantial heterogeneity in the severity analyses (I² of 62% and 86%), and conclude that the evidence is limited by potential biases and heterogeneity across studies, calling for standardised protocols and greater methodological rigour.
One further caution attaches to the headline measure. "Clinical effective rate" is a dichotomised responder outcome — each patient sorted into improved or not — and the threshold for "effective" varies between trials. It is common in this literature and unusually easy to shift in studies where neither the patient nor the practitioner is blinded.
Electroacupuncture, and the same missing comparison
A 2025 multicentre three-arm randomised trial in Complementary Therapies in Medicine tested electroacupuncture, and electroacupuncture combined with warm needling, in older adults with subjective tinnitus.
Both active arms beat the third arm on tinnitus loudness, at both time points:
- Week 5 — electroacupuncture −1.91 (95% CI −2.79 to −1.03); with warm needling −1.81 (−2.68 to −0.94), both p < 0.001
- Week 10 — electroacupuncture −2.03 (−2.95 to −1.11); with warm needling −1.98 (−2.92 to −1.04)
Tinnitus handicap scores improved more in both intervention groups too, the two active arms did not differ from each other, and no serious adverse events occurred.
The third arm was a waitlist. Not sham acupuncture — a group receiving nothing and waiting.
That is the same gap as the scalp acupuncture review above, arriving from the other direction. A waitlist controls for the passage of time and for nothing else: not the attention of a clinician, not the expectation created by a course of treatment, not the effort invested in attending. In a subjective outcome reported by unblinded participants, those are precisely the things a sham is designed to strip out — and most of the sham-controlled trials in the reviews above found no effect. How to read a tinnitus study covers why the control group matters so much.
Taken together, the 2025 evidence on acupuncture for tinnitus is that it performs well against doing nothing and has mostly not performed well against sham needling, the exceptions being two small trials of a single scalp session. Both of those statements have been true for over a decade.
Why the evidence has stayed this murky for so long
A recurring theme across these reviews is methodological weakness in the underlying trials — small sample sizes, inconsistent outcome measures between studies, and varying acupuncture techniques and protocols that make it hard to compare results directly or combine them meaningfully. This isn't unique to tinnitus acupuncture research; broader reviews of acupuncture research quality across many conditions have found similar methodological concerns are common in the field generally.
What this means practically
Acupuncture for tinnitus doesn't have the same clear negative verdict as ginkgo biloba, but it also doesn't have the kind of consistent, well-supported positive evidence that CBT has. It sits in a genuine "we don't know yet" category — not because it's been proven not to work, but because it hasn't been well enough studied to say it does.
Acupuncture from a licensed practitioner is generally low-risk when done properly, which is a meaningful difference from treatments that carry real medical risk despite weak evidence. If you're already inclined to try it, or find it helps with related issues like stress or general well-being, there's little downside beyond cost and time. It's a reasonable thing to try as a complement to better-evidenced approaches — sound therapy, CBT, addressing an underlying cause — rather than as a primary strategy relied on alone. The needle-free variant applied to the outer ear has its own, thinner evidence, covered in ear seeds and auriculotherapy.
Where the professional guidelines land
The American Academy of Otolaryngology–Head and Neck Surgery's clinical practice guideline on tinnitus takes no position on acupuncture — it states no recommendation either way, on the grounds that the evidence was insufficient to support one. The 2024 US Department of Veterans Affairs and Department of Defense guideline lands in the same place: there is insufficient evidence to recommend for or against acupuncture for tinnitus management. The VA/DoD guideline is covered in its own article.
That is a different verdict from the recommendations against supplements, drugs and transcranial magnetic stimulation in the AAO-HNS guideline, and it is worth reading precisely. A panel declining to rule is not a panel finding it useless; it is a panel finding the trials too weak and too heterogeneous to answer the question. The pooled analyses since have not changed that.