OBJECTIVE: The validity of rapid cycling as a distinct course modifier for bipolar disorder was assessed by comparing patients with and without a history of rapid cycling (4 or more affective episodes in 12 months) on demographic, clinical, family history, and outcome variables. These data were also used to formulate operational criteria for the modifier. METHOD: Data on subjects with rapid-cycling (N = 120) and nonrapid-cycling (N = 119) bipolar disorder from four sites were pooled and analyzed by using case-control and historical cohort methods. RESULTS: The rapid-cycling group contained more women and more subjects from higher social classes than the nonrapid-cycling group. Family history did not differ between the groups. The diagnosis had predictive validity in that the rapid-cycling patients had more episodes than the nonrapid-cycling patients during prospective follow- up. The relationship between gender and episode frequency supported the validity of the cutoff point of 4-8 episodes per year. The data regarding whether patients with rapid cycling based on truncated episodes more closely resembled rapid-cycling or nonrapid-cycling patients were equivocal. Patients whose only rapid cycling was associated with antidepressants resembled spontaneously rapid-cycling patients, while the majority of spontaneously rapid-cycling patients also had periods of antidepressant-associated rapid cycling. CONCLUSIONS: The validity of rapid cycling as a distinct course modifier for bipolar disorder is supported by differences in gender, prospectively assessed outcome, and perhaps social class between rapid- cycling and nonrapid-cycling patients. The relationship of gender to episode frequency supports the cutoff of 4 or more episodes per year.Abstract Teaser