New agents with increased activity and/or reduced toxicity are needed for the treatment of advanced breast cancer. The anthracene derivatives mitoxantrone and bisantrene had significant activity and acceptable toxicity in phase II trials. In an ongoing phasc III trial we have now randomized 150 patients with advanced breast cancer to either doxorubicin (60 mg/m2), mitoxantrone (1.4 mg/m 2) or bisantrene (260 mg/m z) i.v. q 3 weeks with re-randomization for cross-over at the time of progression to determine the relative efficacy and toxicity of these three agents. To be eligible, patients must have had only one previous chemotherapy regimen. ER positive patients must have failed endocrine therapy. Patients with CHF or severe cardiac disease were ineligible. In this preliminary evaluation, 117 patients are evaluable for response and 110 for toxicity. Median age for all patients is 58 years (range 26-78). The majority (86%) are postmenopausal. Fifty-nine percent of the patients have visceral dominant disease. Estrogen receptor is positive in 37%, negative in 39% and unknown in 24% of patients. Median performance status (SWOG) is 1, range 0-2. Objective responses have been observed on each arm (doxorubicin 9/35, mitoxantrone 6/38, bisantrene 6/44). Thirty-two paticnts are evaluable for cross-over response (doxorubicin 2/13, mitoxantrone 1/11, bisantrcne 0/8). The predominant toxicity is leukopcnia with a nadir WBC. count <2000 in 45% of all courses administered. Lcukopcnia is similar with the three drugs. Significant nausea, vomiting and aiopecia are common with doxorubicin and uncommon with the other agents. Congestive heart failure has been observed in one patient (doxorubicin). Definitive conclusions regarding the efficacy and toxicity of these agents await the completion of this trial.