Two hundred-two patients with FIGO stages III and IV epithelial ovarian cancer were randomized to 6 or 12 cycles of cyclophosphamide, Adriamycin, and cisplatin (CAP). Patients in complete clinical response underwent a second-look laparotomy, 1 month after cessation of chemotherapy. Patients randomized to 6 cycles and found to be in partial remission at second-look were to receive a further 6 cycles of CAP. Rate of complete pathological response was 23% for 6 cycles of CAP and 25% for 12 cycles; the median survival was 23 months for 6 cycles and 27 months for 12 cycles, and 3-year survival was 29% for 6 cycles and 35% for 12 cycles. None of these differences were statistically significant. Fifty-four patients randomized to 6 cycles were found to be in partial surgical remission at second-look laparotomy, and 24 of these patients agreed to a further 6 cycles and a third-look laparotomy. Six of these 24 patients had a complete pathological response at third-look, improving the complete response rate to 28% in those originally randomized to 6 cycles. However, 3 of these patients all had macroscopic tumors removed at second-look, and two had microscopic disease at second-look. Among patients achieving complete response mean cumulative doses in the CAP 6 cycle group were approximately 50% of those in the CAP 12 cycle group. However, when all patients were considered, this difference was only approximately 15% owing the continuation of chemotherapy in the partial responders of the 6 cycle group and early stopping for chemotherapy in the CAP 12 cycle group due to toxicity or progression. Patients in complete pathological response also showed similar survivals for 6 and 12 cycles. In conclusion, the study did not show a correlation between mean cumulative doses and complete pathological response and survival.