The data on 914 patients enrolled in four randomised trials in advanced ovarian cancer, consecutively conducted by the same cooperative group between 1978 and 1986, were analysed with the aims of: (1) determining the impact of selected prognostic variables on survival; (2) finding, from the interaction of favourable prognostic factors and treatment, an approximate estimate of the magnitude of the survival advantage associated with the use of platinum-based combination chemotherapy. The overall 3-year survival in this series of patients is twice that reported historically (22%; 95% CL 18.7-25.4). The proportional hazard regression model was used to perform the analysis on survival. Residual tumour size, age, FIGO stage and cell type were all independent determinants of survival. Differences in survival from the various prognostic groups were impressive with 5-year survival rates ranging from 7 to 62%. However, these differences were not qualitative (i.e. the kinetics of survival were similar for the best and the worst groups) suggesting that current prognostic factors are of little use for selecting 'biologically' different sub-populations. Platinum-based regimens were associated to an overall prolonged median survival, but this benefit was not observable in the subgroup with most favourable prognosis (less than 2 cm residual tumour size). The implications of these observations for clinical research and ovarian cancer patients care are discussed.