This study was designed to assess the role of dosage of chemotherapy for treatment of metastatic breast cancer. One hundred thirty-three patients without prior chemotherapy for metastatic disease were randomly allocated to receive two different dose levels of cyclophosphamide (C), methotrexate (M), and fluorouracil (F), administered intravenously (IV) every 3 weeks. Patients were stratified by sites of disease (visceral, bone, or soft-tissue dominant) and by interval from primary surgery to first recurrence. Doses on the higher-dose arm were 600 mg/m2 (C,F) and 40 mg/m2 (M) with escalation if possible; doses on the lower-dose arm were 300 mg/m2 (C,F) and 20 mg/m2 (M) without escalation. Patients who failed to respond to lower-dose CMF were crossed over to the higher-dose arm. Patients randomized to the higher-dose arm had longer survival measured from initiation of chemotherapy (median survival, 15.6 months v 12.8 months, P = .026 by log-rank test), but the effect of dose was of borderline significance (P approximately 0.12) when adjusted for a chance imbalance between the two arms in the time from first relapse to randomization, using the Cox proportional hazards model. Response rates (International Union Against Cancer [UICC] criteria) for patients with measurable disease were higher-dose arm: 16/53 (30%) and lower-dose arm: 6/53 (11%), (P = .03). Only one of 37 patients responded on crossover from the lower- to the higher-dose arm. Patients experienced more vomiting, myelosuppression, conjunctivitis, and alopecia when receiving higher doses of chemotherapy. A series of 34 linear analogue self-assessment scales were used to make detailed quality of life assessments on a subset of 49 patients. These scales confirmed greater toxicity in the immediate posttreatment period, but also a trend to improvement in general health and some disease-related indices, in patients receiving higher-dose chemotherapy. This trial suggests that better palliation is achieved by using full-dose chemotherapy.