Seventy-eight advanced breast cancer patients with hormone-resistant disease or visceral metastases were randomized to receive either of two low dose regimens consisting of cyclophosphamide (C), methotrexate (M), 5-fluorouracil (F), and Adriamycin (A) as their initial chemotherapy. One group was treated with CAMF, and the other with CMF until progression, followed by A (CMF leads to A). C was given at 50 mg/m2, po, days 1-14; M at 20 mg/m2, F at 300 mg/m2, and A at 20 mg/m2, iv, days 1 and 8 of each 28-day cycle. The response rates for CAMF vs. CMF did not differ significantly (complete and partial responses-62% vs. 49%; stabilizations-23% vs. 31%). Responses by site of metasis, median times to progression and median survivals were similar for both groups. Poor and good risk partial responders had similar survivals. Twelve percent of CMF patients treated with Adriamycin at the time of progression had partial responses with an associated improved survival. Since CMF is as effective as CAMF, but has less toxicity, low dose therapy with CMF is more acceptable than CAMF as an initial chemotherapy regimen for metastatic breast cancer. Adriamycin may be reserved for subsequent regression induction.