A total of 55 patients with primary dysmenorrhea who had shown a favorable response to a preliminary treatment cycle with placebo were admitted to a double-blind study on placebo versus antiprostaglandin agents (naproxen and pirprofen). To evaluate the placebo effect and its duration, the treatment was given for 4 successive cycles. Whereas the antiprostaglandin agents were effective in most of the patients (in 80% of the pirprofen group and 85.7% of the naproxen group) and this efficacy was maintained throughout the study, a favorable response to placebo was observed in 84% in the first cycle, 29% in the second, 16% in the third and 10% in the fourth. The incidence of side effects was similar in the placebo and the active treatment groups (35.4% vs. 37.5%). It is postulated that a placebo effect in dysmenorrhea is due to a central analgesic mechanism mediated by endorphin release or possibly to psychological dynamics (mental or conditioning theories). However, this effect loses efficacy with time possibly due to a decreased susceptibility to the opioid action of the central nervous circuits responsible for menstrual pain perception or to deconditioning mechanisms.