This phase II trial was conducted to evaluate the percentage of objective responses and the toxicity profile of combination doxorubicin (Adriamycin) and paclitaxel (Taxol) with granulocyte colony-stimulating factor as first-line therapy for patients with metastatic breast cancer (MBC) not previously exposed to anthracycline-containing regimens. Patients with measurable, visceral-dominant MBC and a performance status of 0 to 2 were included in the study. Doxorubicin 60 mg/m2 was administered as a short intravenous infusion, followed by paclitaxel 250 mg/m2 as a 3-hour intravenous infusion on day 1. Granulocyte colony-stimulating factor 5 micrograms/kg/d was given prophylactically as a subcutaneous injection from day 2 until granulocyte recovery to > or = 1,500/mm3. Treatment was repeated every 21 days for a maximum of six courses. Dose reductions (to doxorubicin 50 mg/m2 and paclitaxel 175 mg/m2) and/or treatment delay were applied in case of severe toxicity. All 25 women who entered were evaluable for response and toxicity. The main grade 3/4 toxicities observed were leukopenia, thrombocytopenia, and mucositis. Alopecia occurred in all patients. No clinically relevant cardiovascular toxicity was observed. Severe myelosuppression and/or mucositis necessitated dose reductions at courses 2 or 3 in all but one patient. The complete response rate was 28%, and the partial response rate was 52% for an overall objective response rate of 80%. Median progression-free survival for complete responders was 11 months (range, 3 to 24 months), while the progression-free survival was 7+ months (range 2 to 14+ months) for partial responders and 5 months (range, 3 to 9 months) for nonresponders. This combination produces a high objective response rate in women with MBC, but dose reductions were necessary in almost all cases. Toxicity was manageable after dose reduction, allowing patients to be re-treated for two to six courses without life-threatening toxicity or toxic deaths. Unfortunately, the duration of response was limited even among complete responders. Further trials of this combination in patients with MBC should explore improvements in this study regimen.