Patients with metastatic breast cancer were randomly assigned to receive as second-line chemotherapy either MMM (mitomycin 8 mg/m2 day 1; mitoxantrone 8 mg/m2 days 1 and 22; methotrexate 35 mg/m2 days 1 and 22) alone or in combination with filgrastim (5 micrograms/kg s.c. days 4-17, 24-37). The courses were repeated every 42 days for a maximum of six courses. Thirty-one patients are evaluable for safety and efficacy. The 16 patients in the filgrastim arm received a total of 42 cycles compared with 34 cycles in the 15 control patients. Tumor responses were few in both patient groups (one partial response in the filgrastim group and two partial responses in control group). Nevertheless, a difference in survival was seen (filgrastim median 10.7 months, control median 6.5 months; p = 0.02 log rank). The treatment was well tolerated. Doses were reduced six times in the filgrastim arm and eleven times in the control arm. Grade IV neutropenia was seen in four patients in the filgrastim arm and in twelve patients in the control arm. The observed survival benefit needs to be confirmed in a larger patient group.