The present study examined the associations between drinking water and urinary arsenic levels and skin lesions among 167 residents of three contiguous villages in Bangladesh. Thirty-six (21.6%) had skin lesions (melanosis, hyperkeratosis, or both), of which 13 (36.1%) occurred in subjects who were currently drinking water containing concentrations of arsenic < 50 micrograms/L. The risk for skin lesions in relation to the exposure estimates based on urinary arsenic was elevated more than 3-fold, with the odds ratios for the highest versus the lowest quartiles being 3.6 (95% confidence interval, 1.2 to 12.1) for urinary total arsenic and 3.2 (95% confidence interval, 1.1 to 10.0) for urinary creatinine-adjusted total arsenic. The risks for skin lesions in relation to the exposure estimates based on arsenic in drinking water were less strongly elevated, with the odds ratios for the highest versus the lowest quartiles of exposure being 1.7 (95% confidence interval, 0.6 to 5.1) for drinking-water arsenic and 2.3 for cumulative arsenic index. The study suggests that arsenic exposure is associated with skin lesions in the Bangladesh population and that urinary arsenic may be a stronger predictor of skin lesions than arsenic in drinking water in this population.