The data from two prospective randomised phase III trials that were initiated by the West Midlands Ovarian Cancer Study Group (WMOCSG) in 1981 and 1986, recruiting 167 and 195 patients respectively, have been pooled and the survival patterns of the 362 patients treated for advanced epithelial ovarian cancer within clinical trials in the West Midlands over the 10 year period (1981-91) have been explored. All patients had histologically proven epithelial ovarian cancer and all had residual disease after primary surgery, with the majority having stage III/IV disease. The primary treatment for all patients was debulking surgery followed by platinum-based chemotherapy. Eligible patients were further randomised to undergo a second debulking operation. The main end point, survival, was assessed using Kaplan-Meier curves and the log-rank test. A Cox proportional hazards model identified performance status (P = 0.002), residual disease (P = 0.005) and albumin level (P = 0.04) as independent prognostic factors. A multivariate model to predict survival curves for patients with the best and worst prognoses was developed with predicted 5 year survival of 30% and 3% for those in the best and worst prognostic groups respectively. The identification of clinical interventions to improve outcome is an urgent matter since the prognosis for patients with advanced ovarian cancer remains poor.