Forty chemotherapy naive patients with metastatic or locally advanced breast cancer were treated in a randomized trial comparing mitozantrone 14 mg/m2 with epirubicin 75 mg/m2 given intravenously at 3-weekly intervals. There was a 40% (95% confidence interval (CI) 8-72; P = 0.013) higher partial response rate with epirubicin (11/18) than with mitozantrone (4/19). Epirubicin caused significantly more alopecia (difference 76%; 95% CI 57-96; P < 0.0001) and nausea/vomiting (difference = 38%; 95% CI 10-67; P = 0.01). Three patients who received long courses of epirubicin experienced cardiac failure; two were proved to have cardiomyopathy. The median survival for the epirubicin and mitozantrone groups were 9.5 and 8 months respectively. Thus, although epirubicin gave a higher response rate it also caused more toxicity.