Nineteen factors were analyzed for prognostic significance in a series of 89 women with advanced (stage III or IV) ovarian carcinoma treated with chemotherapy after initial debulking surgery. Seventy-eight of these women received cyclophosphamide, Adriamycin (Adria Laboratories, Columbus, Ohio), and cisplatin (CAP) treatment, and 11 received cyclophosphamide initially with Adriamycin and cisplatin administered at the time of recurrence. Median survival and remission duration were 25 and 19 months, respectively. Using survival as an end point, significant prognostic factors in univariate analyses included the total residual mass after debulking (P = .0007), largest residual mass after debulking (P = .0008), and stage (P = .0098). Using remission duration as an end point, significant prognostic factors in univariate analyses included total residual mass after debulking (P = .007) and the largest residual mass after debulking (P = .0020). The prognostic variables were then considered as possible predictors of survival in a multivariate analysis using the Cox proportional hazards model resulting in the following expression: lambda i(t)/lambda o(t) = exp(0.5928 (log TRM - 1.8117) + 0.6450 (stage - 0.3827) + 0.6673 (C4 - 0.4198) - 0.8596 (CAP - 0.8642)), where lambda i(t)/lambda o(t) is the risk of dying for a particular patient compared with the average risk of the entire group; log TRM is the log of the volume of the total residual mass in cm3 plus 1.0; stage = 0 if stage III, 1 if stage IV; C4 = 0 if cytologic grade is 1, 2, or 3 and 1 if grade 4; CAP = 0 if treatment is cyclophosphamide and 1 if CAP. Median survival times of patients with relative risk greater than 1 and less than 1 are 43 and 19 months respectively. If this model is confirmed in a prospective study, then it could be used to assign risk and assess treatment options for similar patients at diagnosis.