The combination of cyclophosphamide, doxorubicin, and cisplatin (CAP) was reported to be effective in patients with metastatic transitional cell cancer of the urinary bladder. This study was designed to test the effectiveness of the phase II agent amsacrine (m-AMSA) versus CAP in patients previously untreated with systemic chemotherapy, with crossover if no response occurred or at the time of progression. In 23 patients who were randomized to receive CAP, three achieved complete response, seven achieved partial response, six had stable disease, and seven had disease progression. The responses in the 22 patients who received m-AMSA were: one with complete response, three with partial response, six with stable disease, and 12 with progressive disease. The overall response rate to initial CAP therapy was 43% compared to 19% for initial m-AMSA therapy. The median duration of response to CAP was 28 weeks and to m-AMSA was 21 weeks. There were no statistically significant differences in the durations of response or survival times between the groups. The main side effects of CAP were: nausea and/or vomiting, leukopenia, anemia, thrombocytopenia, and renal toxicity. Leukopenia and anemia were the major toxic effects of m-AMSA. Our study supports the justification for testing phase II agent(s) in previously untreated patients with bladder cancer with systemic chemotherapy, provided adequate crossover to a known active single or combination of agents is built into the design of such a trial.