From July 1984 to November 1987, 89 patients with recurrent measurable squamous-cell cancer of the uterine cervix were randomized in a single institution to receive treatment with either carboplatin (CBDCA) or iproplatin (CHIP). Objective response rates were similar: 2 complete regressions (CRs) and 10 partial regressions (PRs) were recorded both in the 46 evaluable patients treated with CBDCA (response rate, 26.1%; 95% confidence interval, 15-41%) and in the 40 evaluable patients treated with CHIP (response rate, 30%; 95% confidence interval, 17-47%). The median duration of response was 5.5 months for CBDCA and 6 months for CHIP; the median survival was 7.5 and 7.6 months, respectively. Both drugs were given in an outpatient setting and myelosuppression (thrombocytopenia) was the predominant toxicity. Analysis of all toxic events yielded additional interesting observations: the occurrence of moderate to severe platelet nadirs beyond cycle 1 was confined to CHIP, a higher incidence of gastrointestinal toxicity during treatment with CHIP, and five moderate to severe complaints of asthenia (recorded as neurologic events) during CHIP therapy versus only one during treatment with CBDCA. Because of its antitumor activity and its toxicologic advantage, a future role for CBDCA in the treatment of cervical cancer appears likely.