A randomised study was started in chemotherapy-naive patients with advanced soft tissue sarcomas who received either epirubicin 60 mg/m2/24 h (total dose for cycle 180 mg/m2) days 1, 2 and 3, (group A) or epirubicin 60 mg/m2/24 h days 1, 2 and 3 and cisplatin 30 mg/m2/24 h days 2, 3, 4 and 5 (group B). The maximal number of cycles foreseen in both groups was eight. Cardiotoxicity of the regimens was monitored by serial LVEF determinations. 106 patients entered this study, 50 (45 evaluable for activity) randomised to group A, and 56 (54 evaluable for activity) to group B. The groups were well balanced for age, sex, performance status and histological type. In group A, there was 1 complete response (CR) and 12 partial responses (PR), the overall response being 13/45 (29%); in group B, there were 7 CRs and 22 PRs, the overall response being 29/54 (54%). The difference between the overall response was statistically significant (chi 2 = 6.19, P < 0.025). The epirubicin-cisplatin regimen was found to be more toxic for platelets and more emetogenic, but cardiotoxicity, either acute or cumulative, was not found to be a major problem in both groups. However, a complete responder receiving a cumulative epirubicin dose of 1440 mg/m2 died from congestive heart failure after a disease-free interval of 27 months. The high response in group B could be the result of the synergism between high-dose epirubicin and cisplatin in patients with advanced soft tissue sarcomas.