Stage III and IV epithelial ovarian cancer patients were prospectively randomized to receive eight courses of 60 mg/m2 of cisplatin plus either 75 mg/m2 of epirubicin (62 patients) or 60 mg/m2 of doxorubicin (54 patients). Clinical response rates for cisplatin/epirubicin of 42% [15% complete response (CR) and 27% partial response (PR)] and for cisplatin/doxorubicin of 55% (24% CR and 31% PR) were not statistically different (p = 0.14). The negative second look rate was 35% (10/29) for cisplatin/doxorubicin and 17% (5/30) for cisplatin/epirubicin (p = 0.12). The progression-free interval for cisplatin/epirubicin (13 months) was not statistically different (p = 0.09) from that for cisplatin/doxorubicin (19 months). The median survivals for cisplatin/epirubicin (756 days) and cisplatin/doxorubicin (739 days) were similar (p = 0.70). Cardiotoxicity was greater for the cisplatin/doxorubicin group (p = 0.0003). With similar survival and less cardiotoxicity, the cisplatin/epirubicin regimen had the more favorable therapeutic index.