The Gynecologic Oncology Group has conducted a randomized prospective trial comparing cisplatin 50 mg/m2 every 21 days (regimen 1), 100 mg/m2 every 21 days (regimen 2), and cisplatin 20 mg/m2 for five consecutive days repeated every 21 days (regimen 3). Four hundred ninety-seven evaluable patients have been accrued on this study. The response rates were 20.7%, 31.4%, and 25.0%, for regimens 1, 2, and 3, respectively; the complete remission rates were 10.0%, 12.7%, and 8.6% for regimens 1, 2, and 3, respectively. The median duration of response ranged from 3.9 to 4.8 months, the median progression-free interval from 3.7 to 4.6 months, and the median survival time from 6.1 to 7.1 months. The difference in response rates for regimens 1 and 2 is statistically significant (P = .015) but less than the magnitude originally considered clinically significant. The differences in complete remission rates, response duration, progression-free interval, and survival times are not statistically significant. The following types of toxicity were observed: serum creatinine level greater than 2 mg/dL and/or BUN level greater than 40 mg/dL was 7%, 14%, and 17% on regimens 1, 2, and 3, respectively; leukocyte count less than 4,000/microL was 27%, 44%, and 41% on regimens 1, 2, and 3, respectively. Nausea and vomiting occurred in 74 patients (83%). The regimen consisting of a 100-mg/m2 single dose has produced a statistically significant higher response rate than the 50 mg/m2 regimen while producing no appreciable differences in complete remission rate, response duration, progression-free interval, or survival. In addition, the higher dose regimen was associated with greater myelosuppression and nephrotoxicity.