This is the first randomized phase II/III trial comparing two carboplatin-based chemotherapy regimens in patients with urothelial cancer who are ineligible ("unfit") for cisplatin chemotherapy. The primary objective of the phase III part of this study was to compare the overall survival (OS) of chemotherapy-naive patients with measurable disease and an impaired renal function (glomerular filtration rate < 60 but > 30 mL/min) and/or performance score of 2 who were randomly assigned to receive either gemcitabine/carboplatin (GC) or methotrexate/carboplatin/vinblastine (M-CAVI). To detect an increase of 50% in median survival with GC compared with M-CAVI (13.5 v 9 months) based on a two-sided log-rank test at error rates α = .05 and β = .20, 225 patients were required. Secondary end points were overall response rate (ORR), progression-free survival (PFS), toxicity, and quality of life. In all, 238 patients were randomly assigned by 29 institutions over a period of 7 years. The median follow-up was 4.5 years. Best ORRs were 41.2% (36.1% confirmed response) for patients receiving GC versus 30.3% (21.0% confirmed response) for patients receiving M-CAVI (P = .08). Median OS was 9.3 months in the GC arm and 8.1 months in the M-CAVI arm (P = .64). There was no difference in PFS (P = .78) between the two arms. Severe acute toxicity (death, grade 4 thrombocytopenia with bleeding, grade 3 or 4 renal toxicity, neutropenic fever, or mucositis) was observed in 9.3% of patients receiving GC and 21.2% of patients receiving M-CAVI. There were no significant differences in efficacy between the two treatment groups. The incidence of severe acute toxicities was higher for those receiving M-CAVI.