Ninety-one patients with untreated epithelial ovarian cancer, stages III and IV, were treated according to a therapeutic protocol including cytoreductive surgery whenever possible, chemotherapy with CAP (cyclophosphamide, doxorubicin, and cisplatin) and second-look laparotomy for those patients achieving a clinical remission. Optimal cytoreductive surgery (residual tumor < 2 cm) was not performed in 66 patients (72.5%). A negative second-look laparotomy demonstrated a pathological complete remission in 26 patients (28.5%). After a median follow-up of 80 months, the disease-free survival is 19.7% (18 of 91 patients). Median survival was greater in optimal cytoreductive surgery patients (47 months) than in the rest of the patients (22 months) (P = 0.0000). Survival was also better in pathological complete remission patients (46 months) than in partial remission (PR) or no response patients (22 months) (P = 0.0001). Optimal secondary cytoreductive surgery was possible in 11 patients in PR after chemotherapy. Survival in this group was similar to that of pathological complete remission cases. Currently, 53% of patients with initial residual tumor < 2 cm and complete response at second-look remain free of disease. In a multivariate analysis, residual tumor > 2 cm and stage IV disease were the most significant prognostic factors. The same analysis indicates that response to chemotherapy at second laparotomy is not an independent prognostic factor. In conclusion, our study indicates that the two most important prognostic factors in advanced ovarian carcinoma are the extent of the initial surgery and stage.