To investigate whether danazol is more effective than placebo for the treatment of premenstrual syndrome (PMS), we conducted a randomized, double-blind, crossover study comparing three successive cycles of danazol (200 mg bid) to three cycles of placebo. Thirty-one women meeting rigorous criteria for a diagnosis of severe PMS over two pretreatment cycles were enrolled; 28 of these subjects completed at least one cycle of treatment with symptom recordings, which were entered into the analysis. A significant period effect confounded the planned within-subject analysis and therefore, the main treatment comparisons were confined to the first period only. Symptom scores on the Premenstrual Tension Self-Rating Scale (PMTS), Beck Depression Inventory (BDI), and a Visual Analogue Scale (VAS) were compared for the premenstrual week in the last cycle of treatment. For the 16 patients on danazol, scores on the PMTS decreased by an average of 14.0 (10.7) (standard deviation) points from a baseline of 25.4 (5.6) points. For the 12 patients on placebo, PMTS scores decreased by an average of 3.6 (9.5) points from a baseline of 23.5 (5.8) points (14.0 vs. 3.6; p = .0133, unpaired t-test). Seven (43.8%) of the subjects on danazol achieved a clinically relevant reduction of symptoms into the asymptomatic range (PMTS scores < or = 5) as compared to one (8.3%) of the subjects on placebo. Thus, danazol (200 mg bid) provided greater relief from severe PMS during the premenstrual week than did placebo.