Bipolars treated with electroconvulsive therapy (ECT) during the index episode were matched on the variables of age, sex, previous admissions and previous hospitalizations with 23 bipolars who did not receive ECT. A similar match was made for 42 unipolars who were under the age of 40 at time of admission. All patients were followed for 5 years. Those patients treated with ECT, both bipolars and unipolars, had the same numbers of episodes in follow-up as their matched groups. However, in both bipolar and unipolar ECT-treated patients, there were more follow-up rehospitalizations. The reason for this is not known but three possibilities exist. Successful treatment with ECT may make the family and patient more prone to consider rehospitalization. Secondly, the originally treated ECT patients may have had more aggressive doctors who were more likely to rehospitalize. Finally, ECT may change the course of an individual's illness in such a way that more severe episodes occur and rehospitalizations are necessary. The findings suggest the need for long-term studies following ECT on clinical and biological variables.