A total of 51 patients with locally advanced transitional cell carcinoma of the bladder were entered into a phase II study to evaluate the effect of initial cisplatin (100 mg/m2, Day 2) combined with methotrexate (30 mg/m2, Days 1 and 8) and vinblastine (4 mg/m2, Days 1 and 8). Of 44 evaluable patients, 25 (57%) achieved an objective response of their primary bladder tumour (complete response (CR): 4 patients; partial response (PR): 21 patients); 30 of the 36 patients with micturition disturbances obtained relief. Toxicity was acceptable, but dose modifications due to haematological or renal toxicity were required in 153 of the 690 drug courses. Non-protocol dose modifications were performed 45 times. The experience gained in this trial emphasises the need for improved co-operation between urologists, radiologists and clinical oncologists in such multicentre phase II studies. Difficulties were observed in adhering to dose modification rules, assessing pre-treatment tumour size (due to variation in quality of measure and in timing relative to initial transurethral resection (TUR)) and defining response to chemotherapy (due to failure to perform mandatory investigations at the right time). This study also illustrates that the accepted criteria for assessing response in the primary bladder tumour are difficult to apply in practice and that the results of any study using this indicator lesion require caution in their interpretation.