Seventy-seven patients with progressive metastatic breast cancer refractory to prior therapy participated in a prospective randomized trial designed to compare the efficacy and toxicity of doxorubicin and epirubicin administered as single agents. In arm 1, 60 mg/m2 of doxorubicin and, in arm 2, 90 mg/m2 of epirubicin were administered by 48-h continuous i.v. infusion every 3 weeks. In arm 3, 90 mg/m2 of epirubicin was administered by bolus every 3 weeks. Patients in the three groups had similar characteristics, except that in arm 3 more patients were premenopausal, had more extensive disease, and fewer patients had been exposed to doxorubicin. Objective remission rates were 29, 26, and 13%, respectively for the three arms. Median response durations ranged from 4-6 months. No significant differences occurred in response rate, remission duration, or survival among patients in the three arms. The incidence of gastrointestinal toxicity and alopecia was evenly distributed. Hematologic toxicity was more severe in arms 2 and 3, and there was a higher incidence of infectious complications in arms 2 and 3 compared to arm 1 (p = 0.05). Two episodes of congestive heart failure occurred in arm 1, one in arm 2, and three in arm 3. Although the total cumulative anthracycline dosage was highest in the arm 2 group, they had the lowest incidence of cardiac toxicity. Epirubicin by bolus and doxorubicin administered by continuous infusion have similar potential for cardiac toxicity. Epirubicin administered by continuous infusion appears less cardiotoxic than doxorubicin by either method of administration or epirubicin given by bolus. Epirubicin appears equally active and less cardiotoxic than the parent compound doxorubicin in patients with metastatic breast cancer.