The prognosis of advanced-stage bladder cancer is poor. Chemotherapy, particularly regimens including platinum salts, appears to increase survival moderately but at the cost of severe, mainly renal toxicity. Platinum is a major factor in this toxicity, and new platinum salts (chiefly carboplatin) have therefore been developed. Carboplatin has no renal toxicity at usual doses, and its use does not require concomitant hyperhydration. Its gastrointestinal, otologic, and general tolerability is excellent. In contrast, most patients develop thrombocytopenia, which can be important, but which is always transitory. The platelet count reaches its nadir (grade 2 or 3) at around day 20, and the leukocyte nadir (grade 2 or 3) occurs about day 19. Anemia is rare. The literature on the use of carboplatin for the treatment of advanced-stage urothelial tumors is reviewed. Carboplatin is used at doses varying between 200 and 400 mg/m2, administered in 28-day courses. Dose adjustment is based on serum creatinine level, creatinine clearance, nadir blood cell levels, or previous treatment, reflecting the wide disparity between different studies. Used alone, carboplatin achieved objective responses (ORs) in 14% of patients (3% complete responses, CRs, and 11% partial responses, PRs) in a total group of 327 patients included in 13 trials. In polychemotherapy various combinations of carboplatin with other agents have been reported, most frequently carbopla-tin/methotrexate/vinblastine; the OR rate was 63% (CR rate 19% and PR rate 44%) among 88 patients in four studies. These results confirm the relative efficacy of carboplatin in the treatment of advanced-stage urothelial tumors, particularly when it is combined with other agents. Its efficacy is similar to that of cisplatin, but it is far less toxic. A prospective, comparative trial will be necessary to confirm these data. The pharmacokinetic behaviors of the two platinum salts are markedly different, as carboplatin does not undergo tubular metabolism. The efficacy of carboplatin could be optimized by adapting the dosage to the glomerular filtration rate, which is a more accurate method than extrapolation from the serum creatinine or creatinine clearance values. This has been shown in the case of nonseminomatous germ cell tumors. Calculation of the optimum carboplatin dose should now be applied to urothelial tumors. The general and renal tolerability of a platinum salt is an important element of choice when the efficacies are equivalent. These considerations fully warrant further clinical trials of carboplatin.