Forty-eight women with premenstrual problems were recruited for therapy with either medroxyprogesterone (MPA) or norethisterone (NET), both compared with placebo, in a double-blind cross-over study. Thirty-five (73%) completed the study. At an oral dose of 15 mg daily for 21 days each cycle, both MPA and NET suppressed ovulation, although reduction of urinary total oestrogen excretion was significantly greater with NET. Breakthrough bleeding occurred in 74% of the cycles treated with MPA but only in 22% of those with NET. Symptoms were monitored daily by visual analogue scales. Both progestins significantly reduced breast discomfort, compared with placebo. While MPA significantly improved individual psychological symptom scores by the second active treatment cycle and pooled psychological symptom scores in both active cycles, NET was no more effective than the placebo. Similar numbers from both groups withdrew because of adverse effects. Among the women treated with MPA, the response to active and placebo therapy was related to the pretreatment psychological symptom profile. The results suggest that the beneficial effect of therapy with MPA in women with premenstrual problems was a consequence of disruption of menstrual cyclicity rather than a result of ovulation suppression.