The relationship between the type of cigarette smoked and bladder cancer risk was examined in data from a hospital-based case-control study. Information on smoking, including filtered and nonfiltered cigarette use, was available for 899 male and 427 female patients and 2713 male and 1243 female age-matched controls interviewed in 20 hospitals from nine cities in the US between 1969 and 1984. Smokers of filtered cigarettes had a slightly reduced risk of bladder cancer relative to smokers of nonfiltered cigarettes (odds ratio 0.64, 95% confidence interval 0.38 to 1.10 among male smokers; odds ratio 0.74, 95% confidence interval 0.37 to 1.48 among female smokers). In both sexes there was no statistically significant reduction in risk among switchers from nonfiltered to filtered cigarettes compared with lifetime nonfiltered cigarette smokers (odds ratio 0.93, 95% confidence interval 0.71 to 1.24 among male smokers; odds ratio 0.76, 95% confidence interval 0.40 to 1.46 among female smokers). The causal implications of these findings, in particular a possible initiation effect of tobacco smoke on bladder carcinogenesis, are discussed.