Fifty-eight patients with metastatic breast cancer were randomly treated with a combination of cyclophosphamide 500 mg/m2 on days 1 and 2 plus epidoxorubicin 90 mg/m2 on day 1 every 3 weeks (group A = 18 patients), or paclitaxel 175 mg/m2 cycle plus doxorubicin 50 mg/m2/cycle every 3 weeks (group B = 20 patients), or paclitaxel as above plus epidoxorubicin 90 mg/m2/cycle every 21 days (group C = 20 patients). The trial was designed as a randomized, multi-institutional phase II study where the cyclophosphamide/epidoxorubicin regimen represented the calibration arm. The overall response rate was 50% (95% CL 26-74%) for arm A, 65% (95% CL 41-85%) for arm B and 70% (95% CL 46-88%) for arm C. The complete response rate was 6% for arm A, 10% for arm B and 15% for arm C. Although this trial was non comparative, the median duration of response and median overall survival were almost superimposable in all arms. The taxane-based regimens were associated with significant neurotoxicity in nearly 20% of cases, while febrile neutropenia represented the most severe side-effect. Our data suggest that the anthracycline/taxane combinations are more effective than the epidoxorubicin/cyclophosphamide regimen, at least in terms of objective response rates. These regimens may represent the treatment of choice when oncologists are faced with aggressive visceral metastatic breast cancer in non elderly women.