Non-drug therapies for women with primary dysmenorrhea are primarily based on anecdotal evidence and small-scale clinical studies. This randomized, observer-blinded, clinical trial evaluated the efficacy of spinal manipulative therapy (SMT) in the treatment of women with primary dysmenorrhea. Women were recruited from the Chicago metropolitan area and evaluated for inclusion through four screening levels. One hundred thirty eight women, ages 18-45, with primary dysmenorrhea diagnosed by participating gynecologists, were randomly assigned to either SMT or a low-force mimic (LFM) maneuver. No treatment occurred at menstrual cycle 1. Treatment for both groups took place on day 1 of cycles 2, 3 and 4, and prophylactic treatment of three visits took place during the 7 days before cycles 3 and 4. Main outcome measures were the Visual Analog Scale (VAS) and plasma concentration of the prostaglandin F2alpha metabolite, 15-keto-13,14-dihydro-prostaglandin F2alpha (KDPGF2alpha), measured 15 min before treatment and 60 min after treatment on day 1 of four consecutive menstrual cycles. The Moos' Menstrual Distress Questionnaire (MDQ) was also administered after treatment on day 1 of each cycle. At cycle 2, the post-treatment VAS scores decreased for both groups, with no statistically significant difference in pre- to post-treatment scores between the two groups (P = 0.44). The changes in pre- to post-treatment KDPGF2alpha levels were not statistically different between the SMT and LFM groups (P = 0.15). No treatment effects were detected over the three cycles for VAS, KDPGF2alpha or MDQ (P = 0.65, P = 0.61 and P = 0.78, respectively). However, there were statistically significant linear time effects for VAS (P = 0.008), MDQ (P < 0.001), and borderline significance for KDPGF2alpha (P = 0.054); these decreases were not considered clinically meaningful. The LFM maneuver used in this study was designed to act as a 'placebo-like' control treatment in comparison with SMT. Although it is possible that the trial did not continue long enough for any placebo effect of the LFM to wash out, it seems more likely that this maneuver was indistinguishable from SMT. Therefore, the postulated superior benefit of high-velocity, short-lever, low-amplitude, high-force spinal manipulation to a low-force maneuver is not supported by the results of this study. 1999 International Association for the Study of Pain.