Acupuncture has been studied extensively, in trials of reasonable quality, across many conditions. The findings resist simple summary, which is why both proponents and critics can cite the literature.
The consistent finding
Across many trials and conditions, a pattern recurs.
Acupuncture generally outperforms no treatment or usual care.
It generally does not clearly outperform sham acupuncture — needling at non-traditional points, shallow needling, or retractable needles that don't penetrate.
The gap between real and sham is small and inconsistent across studies.
That combination is the crux. Something is happening; the specific theory does not appear to be what's producing it.
What follows from it
Two readings, both defensible.
The sceptical reading. If needle placement doesn't matter, then the theoretical framework — meridians, specific points, energy flow — is not doing the work. The effect is contextual: attention, ritual, expectation, and the physical experience of the procedure.
The alternative reading. Sham acupuncture may not be an inert control. Inserting needles anywhere produces physiological effects — local tissue response, release of endogenous opioids, effects on nerve signalling.
If both conditions are producing real physiological effects, then finding no difference between them doesn't demonstrate that neither works.
This is a genuine methodological problem and it's why the debate persists rather than resolving.
Where guidelines have landed
Various clinical guidelines have reached different conclusions, which reflects the ambiguity.
Some guidelines recommend consideration of acupuncture for chronic pain conditions, particularly chronic primary pain and certain headache types.
Others have removed previous recommendations after reassessment, or recommend against it on the grounds that effects are not clinically meaningful relative to sham.
The variation between guideline bodies examining broadly the same evidence is instructive about how ambiguous findings get interpreted.
The conditions with better evidence
Where the literature is relatively more supportive.
Chronic pain conditions, particularly musculoskeletal pain and headache, where meta-analyses of high-quality trials have found effects that exceed sham by small margins and that persist over some months.
Chemotherapy-induced nausea, where the evidence is comparatively stronger.
Where the evidence is weaker or absent: most conditions outside pain and nausea, despite claims made for a very wide range.
The mechanism research
Proposed physiological mechanisms have been investigated.
Needle insertion triggers local tissue responses and stimulates nerve fibres, with documented effects including release of endogenous opioids.
Imaging studies have found brain activity changes during acupuncture, though similar changes occur with sham.
Effects on connective tissue have been proposed following observations of tissue response to needle rotation.
None of these mechanisms requires or supports the traditional theoretical framework. They describe effects of inserting needles into tissue, which would apply regardless of point selection.
Safety
Generally good when practised by trained practitioners using single-use needles.
Serious adverse events are rare and documented, including pneumothorax from needling near the lungs and infections from inadequate hygiene.
Minor effects — bruising, bleeding, temporary soreness, occasional fainting — are common and self-limiting.
Regulation varies substantially between jurisdictions, and checking practitioner registration matters.
The practical position
For someone considering it for chronic pain.
The evidence suggests you may experience improvement, that the improvement is real, and that it may not be attributable to the specific technique.
For a condition where conventional options are limited and side effects are a concern, that may be an acceptable proposition. Chronic pain is precisely such a condition.
It should not replace investigation of an undiagnosed problem, and it should not replace treatment for a condition with effective treatment available.
And claims for conditions outside pain and nausea should be treated with considerably more scepticism than the practice as a whole warrants.
The wider point
This case illustrates something general about evaluating practices.
A practice can produce real benefit while its explanatory framework is wrong. Those are separable questions and they get conflated in both directions — defenders citing benefit as evidence for the theory, critics citing the theory's implausibility as evidence against benefit.
The more useful questions are: does it help, compared with what, at what cost, with what risk, and is anything better available.
Those can be answered without settling whether the traditional theory is correct, which is convenient because that question probably cannot be settled.
General information only. Discuss any treatment with a qualified healthcare professional, and ensure any practitioner is appropriately registered.
The dry needling comparison
An interesting parallel that illuminates the mechanism question. Dry needling, used by some physiotherapists, involves inserting needles into muscular trigger points, with a framing based on musculoskeletal anatomy rather than on meridians.
The physical procedure is broadly similar. The theoretical framework is entirely different.
Evidence for dry needling in musculoskeletal pain is comparable to that for acupuncture — modest effects, difficulty establishing superiority over sham.
Which is consistent with the reading that inserting needles into tissue produces effects, and that the theoretical framework attached to it is doing less work than either tradition claims.