Two Adriamycin (doxorubicin)-based chemotherapy regimens were investigated in patients with carcinoma of the breast who had failed prior systemic therapy. The two chemotherapy programs, dibromodulcitol, Adriamycin, vincristine, and Halotestin (fluoxymesterone) (DAVH), and thiotepa, Adriamycin, vinblastine, and Halotestin (TAVH), were chosen for comparison on the basis of reported response rates of 40% to 50% with remission durations of 11 months in patients refractory to other cytotoxic chemotherapy. Cycles of DAVH were repeated every 4 weeks. Cycles of TAVH were repeated every 3 weeks. Of 184 patients evaluable for response, 32% of patients treated with DAVH and 38% of patients treated with TAVH had a complete response (CR) or partial response (PR). An additional 5% of patients had nonmeasurable improvement in osseous disease for an overall rate of response (CR + PR + improvement) of 40%. Patients who had previously received cytotoxic chemotherapy for metastatic disease or had early failure after adjuvant therapy had a lower response rate to DAVH, but not to TAVH than those who did not fail prior chemotherapy. Duration of response and survival were similar with the two treatments. There were seven treatment-related deaths, five among patients receiving DAVH and two among patients receiving TAVH. Patients receiving DAVH had significantly more thrombocytopenia and neurologic toxicity than those receiving TAVH. These treatments appear to be reasonable second-line regimens and are good candidates to be used in initial therapy of metastatic disease or adjuvant therapy studies that explore the use of alternating non-cross-resistant combinations with cyclophosphamide, methotrexate, and 5-fluorouracil.