To evaluate the substitution of methotrexate with doxorubicin (Dox) in CMF-(cyclophosphamide, methotrexate, 5-fluorouracil) containing regimen for advanced breast cancer, 415 postmenopausal patients below the age of 66 years, naïve to chemotherapy, were accrued from 1980 to 1984 and followed-up until 1995. They received tamoxifen 30 mg daily orally and by randomisation either 400 mg/m2, cyclophosphamide, 25 mg/m2 doxorubicin and 500 mg/m2 5-fluorouracil (CAF) or 40 mg/m2 methotrexate instead of Dox (CMF) intravenously (i.v.) days 1 + 8 repeated every 4 weeks. Dox was substituted by methotrexate at a cumulative dose of 550 mg/m2. Among 341 eligible patients the response rate and median time to progression was significantly in favour of CAF: 53% CAF versus 36% CMF (P = 0.002) and 11.8 months CAF versus 6.5 months CMF (P = 0.001). Median duration of response was 19.5 CAF versus 18.0 CMF months, and survival 20.8 CAF versus 17.4 CMF months (non-significant). The two regimens were equimyelotoxic. There were no treatment-related fatalities but 1 patient with congestive heart failure on CAF was reported. Nausea/vomiting, stomatitis and infections were modest in both groups, whilst alopecia was more common with CAF. Regression analysis showed that long recurrence free interval, good performance status, and no visceral involvement was significantly related to long-term survival, whilst the treatment regimen was not. It is concluded that in chemotherapy-naïve patients with advanced breast cancer Dox-containing regimens are superior and remain the first choice of chemotherapy, especially in patients with visceral metastases, until newer drugs and combinations have been proven to be superior.