We tried to determine the role of cytoreductive surgery for stage IV epithelial ovarian cancer and in what conditions this surgical procedure could carry the best benefits. From January 1986 to December 1997, seventy-one of 73 patients with stage IV epithelial ovarian cancer who were treated in Cancer Hospital of Shanghai Medical University were retrospectively reviewed. Clinical information including age, grade, histology, presence of ascites, size of residual disease, site of extra-abdominal metastasis, whether initially presenting as metastatic disease or not, neo-adjuvant chemotherapy, platinum-based chemotherapy and second-line chemotherapy was obtained. Survival was calculated by life-table and survival curves were computed using the Kaplan-Meier method with differences in survival estimated by log-rank test. Independent prognostic factors were identified by Cox's proportional hazards regression model. The median age of the patients' population was 54 years (range 22-82), median follow-up time was 12 months (range 3 to 130) and estimated 5-year survival rate 6.1%. Thirty out of 71 (42.3%) patients were successfully debulked (< or = 1 cm) at the time of initial surgery. There was a significant difference in five-year survival rate between patients optimally (14.1%) vs suboptimally (0%) cytoreduced, with an estimated median survival in the optimal group of 23 months vs 9 months in the suboptimal group (P=0.0001, long-rank test). When the variables were factorized, only in patients with malignant pleural effusion or positive supraclavicular lymph nodes, optimal cytoreduction could get the greatest benefits. Multivariate analysis revealed that the size of residual disease and ascites were independent factors of survival. However, only ascites was the prognostic factor of progression-free survival. Optimal cytoreductive surgery is an important determinant of survival in women with stage IV epithelial ovarian cancer, mainly in those with malignant pleural effusion or positive supraclavicular lymph node pathology.