One hundred ninety-eight postmenopausal women with metastatic breast carcinoma were entered in this study. After six induction cycles with cyclophosphamide, methotrexate and 5-fluorouracil (CMF), patients with at least stable disease were randomized to the "continuation arm" (continuation of CMF until progression) (A, 49 evaluable patients) or to the "intensification-discontinuation arm" (addition of adriamycin and vincristine to two of the three drugs of CMF for six more cycles; i.e., CMAV, CFAV, MFAV, twice; discontinuation of chemotherapy; radiotherapy to pre-study sites of disease in patients prospectively considered as candidates to receive this treatment) (B, 46 evaluable patients). After randomization, escalation of response category occurred in five patients on A (10%) and in five on B (11%). Time to progression was transiently delayed in arm B within 6 months after randomization. There were no significant differences in the overall time to progression, duration of response or survival. On arm B, after discontinuation of chemotherapy, median time to relapse was 22 weeks. This time was significantly longer in patients who were candidates for radiation therapy (36 weeks, P = 0.005), or with a disease-free interval greater than 1 year (32 weeks, P = 0.004) or who achieved complete remission (60 weeks, P = 0.0001). On arm B, three patients (7%) are still alive in complete remission in excess of three, five and six years following discontinuation of therapy. This study indicates that late intensification of chemotherapy followed by discontinuation of treatment may maintain palliation, and allow a long treatment-free period in responding patients with advanced breast carcinoma.(ABSTRACT TRUNCATED AT 250 WORDS)