Sixty patients with metastatic breast cancer refractory to prior doxorubicin combinations were randomized by performance status, dominant disease site, and number of involved organ sites to receive vindesine either as a bolus injection of 3-4 mg/m2 iv every 10-14 days or as a continuous 5-day infusion of 1-1.2 mg/m2/day every 21 days. There were two patients with partial responses (7%) and six with stable disease among the 26 evaluable patients who received bolus injections. Of the 25 evaluable patients who received continuous infusions, seven achieved a partial response (28%) and 11 had stable disease (0.001 less than P less than 0.005). Thirteen patients, after failing to respond to bolus vindesine, were given continuous infusions. Of these, 11 were evaluable and four had partial response (36%). Responses were seen in all organ sites of involvement, with response duration ranging from 2 to 9+ months. Side effects included nausea, vomiting, stomatitis, constipation, neuropathy, fever, and myelosuppression. Except for myelosuppression, which was more evident with the continuous infusion schedule, no significant difference was seen in the frequency of side effects encountered with the two schedules. These results confirmed that there is an improved therapeutic index for vindesine when it is given as a continuous 5-day infusion.