One hundred and thirty-eight patients with stage III and 27 with stage IV ovarian malignant tumors were registered during 1969-1991. Seventy-seven patients deemed resectable received primary debulking followed by postoperative (PO) chemotherapy. Eighty-two patients considered to be unsuitable for curative surgery underwent an exploratory laparotomy, of whom 74 received neoadjuvant (NA) chemotherapy and the remaining 8 did not receive planned NA chemotherapy because of their poor performance status (PS). Therefore, the patients in the NA group had a more aggressive disease and had a poorer PS than those in the PO group. Thirty-five patients in the PO group received optimal debulking (OD: residuum < 2cm) at initial laparotomy, indicating that the OD rate for the initial surgical attempt for stage III and IV disease is 21% (35/165). The response to chemotherapy by the PO and NA groups was 42% and 68%, respectively. Among 74 patients receiving NA chemotherapy, 34(46%) had OD. Survival (5 year, median) for each subgroup was 42%, 38 mo for OD-->PO (35 cases), 6%, 18 mo for subOD-->PO (42 cases), 66%, 82 mo for NA-->OD (34 cases), and 7%, 18 mo for NA-->subOD (40 cases). In all, 19% 21 mo for the PO group and 34%, 31 mo for the NA group. Survival for NA-->OD was significantly higher than that for OD-->PO and almost equal to that for stage II (54%, 72 mo). The present study with a selection bias set at the initial laparotomy clearly indicates that NA chemotherapy followed by OD is a recommendable treatment for improving the prognosis of advanced ovarian cancer.