One hundred eight patients with recurrent or metastatic transitional cell carcinoma of the urothelial tract were randomized to receive cisplatin (C) 80 mg/m2 on day 1 every 4 weeks, or methotrexate (M) 50 mg/m2 on days 1 and 15 plus C 80 mg/m2 on day 2 every 4 weeks (C + M). Fifty-three eligible patients were randomized to C + M and 55 to C. In the C + M arm, 45% of patients responded (complete response [CR], 9%) and 31% (CR, 9%) in the C arm (P = .18). In the C arm, 20 patients failing or relapsing after C received M. Two patients responded, and four with progressive disease (PD) and one with a previous partial response (PR) showed no change. The median survival was 8.7 months (C + M arm) and 7.2 months (C arm), P = .7. Relapse-free survival was not significantly different, but C + M was associated with a significantly increased time to disease progression (median, 5.0 months, v 2.8 months for C arm). The response of untreated patients (37%) was not different from those with prior treatment (39%). On the C + M arm, 92% of patients and 96% of patients on the C arm received 85% or more of the scheduled C dose. Significantly more grade 3 or 4 hematological toxicity (27% v 2%; P = .01) and mucositis (20% v 0%; P = .0005) occurred in patients on the C + M arm. Although the initial response rates seen on the combination arm look superior, and the time to disease progression is increased, these effects have not translated into a clinically important increase in the duration of survival and were associated with increased toxicity.