The carboplatin-based chemotherapeutic regimen M-CAVI (methotrexate, carboplatin, and vinblastine) is active against bladder carcinoma and can be administered to patients who are ineligible to receive cisplatin or doxorubicin. The authors designed a randomized study to evaluate whether M-CAVI offers a therapeutic advantage over the cisplatin-based regimen M-VAC (methotrexate, vinblastine, doxorubicin, and cisplatin) in the treatment of patients with surgically incurable advanced bladder carcinoma. Patients with surgically incurable advanced bladder carcinoma were enrolled on a randomized trial comparing M-CAVI, which consists of carboplatin (300 mg/m2 on Day 2, adjusted using Calvert's formula for an area under the curve of 5), methotrexate (30 mg/m2 on Days 1, 15, and 22), and vinblastine (3 mg/m2 on Days 2, 15, and 22) administered every 28 days, versus standard M-VAC. The eligibility criteria included histologically proven bladder carcinoma, surgically incurable disease, and no prior chemotherapy. Patients were treated until disease progression or unacceptable toxicity occurred. From January 1989 to January 1994, 47 assessable patients were included. Seventeen patients had lymph node disease and 30 had distant metastatic disease. Twenty-three patients were randomized to receive M-CAVI and 24 to receive M-VAC. Patient characteristics in the two groups were similar. Overall response rates were 39% (95% confidence interval [CI], 20-62%) for M-CAVI and 52% (95% CI, 30-73%) for M-VAC (P = 0.3), with 3 complete responses observed among patients treated with M-VAC and none among those in the M-CAVI group. M-VAC was associated with more gastrointestinal toxicity, stomatitis, alopecia, and Grade 4 neutropenia than M-CAVI. One toxicity-related death occurred in the M-VAC group. There was a statistically significant difference in median disease-related survival time favoring M-VAC (16 months; range, 6 to 22+) versus M-CAVI (9 months; range, 6 to 14+) (P = 0.03). M-CAVI is less toxic but less active than M-VAC in the treatment of patients with advanced bladder carcinoma. Carboplatin-based regimens in which carboplatin is administered at the dose range used in the current study should be reserved for patients who cannot tolerate cisplatin treatment. Further research is required to assess the impact of high dose carboplatin in the treatment of this disease.