BACKGROUND To evaluate the antitumor activity in patients with T3b, T4 or metastatic urothelial carcinoma treated with MEC or M-VAC chemotherapy, by performing a multi-center randomized prospective study. METHODS From 1991 to 1995, 89 patients with T3b, T4 or metastatic urothelial carcinoma were randomly allocated to a methotrexate, epirubicin and cisplatin chemotherapy group (arm 1: S-MEC therapy; n = 29), a dose-intensified MEC therapy combined with G-CSF group (arm 2: I-MEC therapy; n = 30) or a methotrexate, vinblastine, doxorubicin and cisplatin chemotherapy (arm 3: M-VAC therapy; n = 30). At the registration center, the patients were stratified into previously untreated patients and patients with recurrence after radical operation and then randomly allocated to the treatment groups. In each arm, two or more courses of chemotherapy (4-week cycles) were performed. RESULTS Of the 88 eligible patients, four treated with S-MEC therapy and two treated with I-MEC therapy showed CR. The response rates (CR + PR) were 52% (15/29) with S-MEC therapy, 76% (22/29) with I-MEC therapy and 47% (14/30) with M-VAC therapy. The response rate with I-MEC therapy was significantly higher than that with M-VAC therapy (P = 0.02). Although the incidence of leukopenia was low with I-MEC therapy, the incidence of thrombocytopenia was high with this therapy. CONCLUSIONS MEC therapy used in this study is promising in terms of the antitumor effects.