How Acupuncture May Work: Proposed Mechanisms and What Is Known
The main scientific ideas about how acupuncture might act on the body, from adenosine and opioids to nerves and connective tissue, and how much of it is tested in people.
Content last updated . This page has not been medically reviewed.
Start with the honest status
NCCIH says how acupuncture works is not fully understood. [1] Every mechanism below is a proposed explanation, and many were studied in animals. A mechanism found in a mouse is not proof of a clinical effect in people, and a clinical benefit does not depend on a known mechanism. This page is separate from the traditional explanation in terms of qi and meridians, which is a framework from the tradition rather than a physiological finding; see Qi and Acupuncture and traditional frameworks vs biomedical explanations. Evidence labels: preclinical, promising but limited, insufficient evidence.The main ideas, side by side
| Proposed mechanism | What was found | Where tested | Label |
|---|---|---|---|
| Local adenosine release | Adenosine was released during acupuncture in mice, and its pain-blocking effect required adenosine A1 receptors. [2] In human volunteers, adenosine rose at the needled site and stayed up 30 minutes after. [3] | Mice, then a human sampling study | Promising but limited |
| Opioid system | Review: opioid peptides and their receptors in a spinal and brainstem pathway play a pivotal role in acupuncture analgesia. [4] A PET study in fibromyalgia found opioid-receptor binding rose after real acupuncture but not sham. [5] | Mostly animals; one human PET trial | Promising but limited |
| Connective tissue signaling | A hypothesis that needle rotation couples with connective tissue and sends mechanical signals to cells. [6] | Proposal | Insufficient evidence |
| Autonomic nervous system and inflammation | Electroacupuncture activated a vagal pathway that controlled inflammation and rescued mice from induced peritonitis. [7] | Mice | Preclinical |
| Nonspecific effects | Belief, the relationship with the practitioner and other factors may contribute. [1] | Human trials | See sham and placebo controls |
Adenosine
In the mouse study, an adenosine A1 receptor agonist replicated the pain relief of acupuncture, and blocking adenosine breakdown extended it. [2] The human study sampled fluid at the Zusanli point. Adenosine release was not seen if the needle was placed away from that point or inserted without being rotated. [3] This was one lab experiment in volunteers, not a trial of pain relief, so it supports the idea without showing a clinical effect.Opioids and the brain's pain systems
A review of animal and imaging work described electroacupuncture at different frequencies releasing different opioid peptides in the spinal cord, and it said analgesia appeared only when the patient felt the characteristic needling sensations. [4] A separate review of preclinical studies reported that electroacupuncture inhibited inflammatory and neuropathic pain more effectively at 2 to 10 Hz than at 100 Hz. [8] Those are laboratory findings in animals and say nothing yet about what frequency suits any person. See electroacupuncture.Connective tissue
A 2001 paper proposed that "needle grasp" is caused by connective tissue winding around a rotated needle, which could send mechanical signals to cells. [6] It is framed in the paper as a hypothesis. This page reports only that original proposal.Nerves, the autonomic system and inflammation
A mouse study found that electroacupuncture at a hindlimb point, but not an abdominal point, can drive the vagal-adrenal anti-inflammatory axis. [9] This helps explain why some points might act differently from others in animals. It does not show that electroacupuncture treats inflammation or infection in people, and nobody should use it in place of care for infection or sepsis. Another group found that electroacupuncture at the sciatic nerve controlled systemic inflammation in mice. [7]What this means for you
- Mechanism research is active, but it does not tell you whether to try acupuncture. Clinical trials do that: see Does Acupuncture Work?.
- Studies that use imaging add to the picture: acupuncture, brain imaging and neuroscience.
- NCCIH says studies in animals and people, including imaging, have shown acupuncture may affect nervous system function. [1]
Frequently asked questions
Is it known how acupuncture works?
No. NCCIH describes the mechanism as not fully understood, with possible effects on the nervous system, other tissues and nonspecific effects. [1]
Does acupuncture release endorphins?
Animal studies and a human PET study point to the opioid system. [4][5] It remains a proposed mechanism, not an established one.
Do acupuncture points have special properties?
In mice, electroacupuncture at one hindlimb point but not an abdominal point drove an anti-inflammatory pathway. [9] Whether human points differ in the ways the tradition describes has not been shown by the sources we read.
Is adenosine the answer?
It is one candidate. The human study showed a local rise in adenosine, but not that it causes pain relief. [3]
Does a known mechanism prove it works?
No. Only clinical trials in people with a condition can show whether it helps. [1]
More details: evidence, limits and cautions
The details behind this page: cautions, limits of the evidence and where sources disagree. Open any note to read it.
Safety
Mechanism findings do not change safety advice. Tell any practitioner about blood-thinning medicine, a pacemaker or pregnancy, and read side effects and precautions.
Education, not medical advice. This page is for general information. It does not diagnose or treat any condition and is not a substitute for care from a qualified health professional. Natural does not automatically mean safe. Ask a clinician or pharmacist before changing your care, especially if you take medication, are pregnant or breastfeeding, or have a health condition.