Thirty-five patients with advanced breast cancer refractory to initial chemotherapy were randomized to receive two-drug doxorubicin-CCNU or three-drug doxorubicin-CCNU-vincristine (VCR) treatment. Doxorubicin (25-40 mg/m2) and VCR (1.4 mg/m2) were given intravenously every 21 days; CCNU (65-90 mg/m2) was given orally every 42 days. Patients with liver or bone marrow metastases initially received the lower range doses. All patients failed prior chemotherapy with cyclophosphamide and 5-FU with or without methotrexate and prednisone; 67% received prior hormonal therapy. Pretreatment patient characteristics were comparable in both groups. Objective responses (greater than 50% reduction in tumor size) were seen in 7 of 18 patients (39%) on the VCR containing arm and in 8 of 17 patients (46%) on the doxorubicin-CCNU arm. Survival was not improved by VCR addition with overall survival on the doxorubicin-CCNU arm, approximately 10% greater at all intervals (median 9.1 versus 7.0 months; not statistically significant). However, neurotoxicity was significantly greater in the VCR arm (36% versus 0%; P less than 0.05). In advanced breast cancer, no benefit to VCR addition was seen in prior randomized studies of first-line combinations where VCR was the treatment variable. Such results, combined with our experience with VCR in a salvage regimen, question the value of VCR addition to combination regimens in advanced breast cancer.