A total of 205 women with stage III or IV ovarian cancer who had persistent disease after initial treatment with doxorubicin and cisplatin were randomized to receive melphalan (8 mg/m2 orally for 4 days) or the combination of melphalan (6 mg/m2 for 4 days) and hexamethylmelamine (120 mg/m2 for 14 days) every 4 weeks. Only one of 64 patients with measurable disease had an objective response. The major determinants of survival after randomization were the amount of residual disease after initial chemotherapy and the type of response to initial chemotherapy. There was no overall difference in survival between the two chemotherapy regimens, but the small group of patients whose disease progressed on initial chemotherapy survived significantly longer when treated with the two-drug combination. Neither of these regimens provided effective therapy for women whose disease was not eliminated by first-line treatment. However, the superior results obtained in one subgroup with the addition of hexamethylmelamine suggest that the place of this agent in treating ovarian cancer should be carefully evaluated.