Background: Despite substantial increases in its popularity and use, the efficacy of acupuncture for chronic mechanical neck pain remains unproved.

Objective: To compare acupuncture and placebo for neck pain.

Design: A randomized, single-blind, placebo-controlled, parallel-arm trial with 1-year follow-up.

Setting: The outpatient departments of 2 major hospitals in the United Kingdom, 1999 to 2001.

Patients: 135 patients 18 to 80 years of age who had chronic mechanical neck pain. Eleven patients withdrew from treatment, and 124 completed the primary end point.

Measurements: The primary outcome was pain 1 week after treatment, according to a visual analogue scale. Secondary outcomes were pain at other time points, score on the Neck Disability Index and the Short Form-36, and use of analgesic medications.

Interventions: Patients were randomly assigned to receive, over 4 weeks, 8 treatments with acupuncture or with mock transcutaneous electrical stimulation of acupuncture points using a decommissioned electroacupuncture stimulation unit.

Results: Both groups improved statistically from baseline, and acupuncture and placebo had similar credibility. For the primary outcome (weeks 1 to 5), a statistically significant difference in visual analogue scale score in favor of acupuncture (6.3 mm [95% CI, 1.4 to 11.3 mm]; P = 0.01) was observed between the 2 study groups, after adjustment for baseline pain and other covariates. However, this difference was not clinically significant because it demonstrated only a 12% (CI, 3% to 21%) difference between acupuncture and placebo. Secondary outcomes showed a similar pattern.

Limitations: All treatments were provided by 1 practitioner. Although the control was credible, it did not mimic the process of needling. A nonintervention group was not present to control for regression to the mean.

Conclusions: Acupuncture reduced neck pain and produced a statistically, but not clinically, significant effect compared with placebo. The beneficial effects of acupuncture for pain may be due to both nonspecific and specific effects.