The Swedish Tumor Registries are population based. Three hundred eighty-four patients with epithelial ovarian cancer registered in the Tumor Registry of the Southeast Health Care Region during the years 1984 to 1987 were reviewed; 332 patients were eligible for survival analysis. The protocol treatment during this period included primary surgery aimed at tumor reduction followed by combination chemotherapy (anthracyclin + cis-platinum, AP) in FIGO stages IC-IV. All patients treated otherwise were registered as prescribed nonprotocol treatment. An overall corrected 5-year survival of 40% was recorded. However, only 166/325 of the patients (51%) were actually prescribed protocol treatment as defined above. The 5-year survival was 49% for patients prescribed protocol treatment compared to 33% for those prescribed nonprotocol treatment (P < 0.0001). Stages III-IV patients prescribed protocol treatment had 23% 5-year survival compared to 11% for patients prescribed nonprotocol treatment (P < 0.0001). The impact of cis-platinum-based combination chemotherapy was thus not as pronounced as expected from clinical studies since many patients were not considered to be in a good enough general condition to receive cis-platinum. Multivariate Cox analysis of 223 cases showed that age (P < 0.0001), stage (P = 0.0002), grade of differentiation (P = 0.006), and postoperative residual tumor (P = 0.007) were independent prognostic factors. A prognostic index was developed which divided the patients into different risk groups. It was shown that high-risk patients identified by this index were mainly found in the group prescribed nonprotocol treatment and that the prognostic index could not be used to identify patients with a low probability of response among patients prescribed protocol treatment.