Inorganic arsenic is an established cause of lung and skin cancer. Epidemiological evidence from Taiwan suggests that arsenic causes more fatal internal cancers, with the highest relative risks reported for bladder cancer. We conducted a cross-sectional biomarker study in a Chilean male population chronically exposed to high (70 subjects) and low (55 subjects) arsenic levels in their drinking water (average concentrations, 600 and 15 micrograms As/liter, respectively). A fluorescent version of the exfoliated bladder cell micronucleus (MN) assay was used employing fluorescence in situ hybridization with a centromeric probe to identify the presence (MN+) or absence (MN-) of whole chromosomes within micronuclei, thereby determining the mechanism of arsenic-induced genotoxicity in vivo. We divided the study population into quintiles by urinary arsenic levels and found an exposure-dependent increase in micronucleated cell prevalence in quintiles 2-4 (urinary arsenic, 54-729 micrograms/liter). The largest increase appeared when quintile 4 was compared to quintile 1 [prevalence ratio, 3.0; 95% confidence interval (CI), 1.9-4.6]. The prevalence of MN+ increased to 3.1-fold in quintile 4 (95% CI, 1.4-6.6), and the prevalence of MN-increased to 7.5-fold in quintile 3 (95% CI, 2.8-20.3), suggesting that chromosome breakage was the major cause of MN formation. Prevalences of total MN, MN+, and MN- returned to baseline levels in quintile 5 (urinary arsenic, 729-1894 micrograms/liter), perhaps due to cytostasis or cytotoxicity. These results add additional weight to the hypothesis that ingesting arsenic-contaminated water enhances bladder cancer risk and suggest that arsenic induces genetic damage to bladder cells at drinking water levels close to the current United States Maximum Contaminant Level of 50 micrograms/liter for arsenic.