A single case design, with time series analysis, was employed to evaluate the efficacy of acupuncture in the treatment of tension headache. Fourteen patients were given eight weekly treatments, four of true acupuncture and four of sham in random order. Mean pain in medication scores were reduced by 52% and 54% respectively at initial follow-up. Reductions in pain scores of over 50% were achieved by half the patients and the significance of these changes confirmed by time series analysis. The majority of patients maintained their gains at four month follow-up. True acupuncture was shown to be significantly superior to sham, demonstrating a specific therapeutic action, in four patients. In the remainder no difference was observed. Possible mechanisms for these effects are discussed. Acupuncture is a potentially valuable treatment for tension headache but further research is needed.