Data from a population-based case-control interview study of incident bladder cancer in 10 areas of the United States were used to estimate relative risks among white men (2,116 cases, 3,892 controls) and women (689 cases, 1,366 controls) according to beverage intake level and type of water source. Individual year-by-year profiles of water source and treatment were developed by linking lifetime residential information with historical water utility data from an ancillary survey. Risk of bladder cancer increased with intake level of beverages made with tap water. The odds ratio (OR) for the highest vs. lowest quintile of tap water consumption was 1.43 [95% confidence interval (CI) = 1.23, 1.67; chi 2 for trend = 26.3, P less than .001]. The risk gradient with intake was restricted to persons with at least a 40-year exposure to chlorinated surface water and was not found among long-term users of nonchlorinated ground water. The ORs for the highest vs. lowest quintiles of tap water intake were 1.7 and 2.0, respectively, among subjects with 40-59 and greater than or equal to 60 years' exposure. Duration of exposure to chlorinated surface water was associated with bladder cancer risk among women and nonsmokers of both sexes. Among non-smoking respondents with tap water consumption above the population median, the OR increased with exposure duration to a level of 3.1 (CI = 1.3, 7.3; chi 2 for trend = 6.3, P = .01) for greater than or equal to 60 years of residence at places served by chlorinated surface water (vs. non-chlorinated ground water users). These results extend findings of earlier epidemiologic studies and are consistent with environmental chemistry and toxicologic data demonstrating the presence of genotoxic by-products of chlorine disinfection in treated surface waters.