Based on epidemiological data, chronic exposure to high levels of inorganic arsenic in drinking water is carcinogenic to humans, inducing skin, urinary bladder and lung tumors. In vivo, inorganic arsenic is metabolized to organic methylated arsenicals including the highly toxic dimethylarsinous acid (DMA^III) and monomethylarsonous acid (MMA^III). Short-term treatment of rats with 100 μg/g trivalent arsenic (As^III) as sodium arsenite in the diet or in drinking water induced cytotoxicity and necrosis of the urothelial superficial layer, with increased cell proliferation and hyperplasia. The objectives of this study were to determine if these arsenic-induced urothelial effects are dose responsive, the dose of arsenic at which urothelial effects are not detected, and the urinary concentrations of the arsenical metabolites. We treated female F344 rats for 5 weeks with sodium arsenite at dietary doses of 0, 1, 10, 25, 50, and 100 ppm. Cytotoxicity, cell proliferation and hyperplasia of urothelial superficial cells were increased in a dose-responsive manner, with maximum effects found at 50 ppm As^III. There were no effects at 1 ppm As^III. The main urinary arsenical in As^III-treated rats was the organic arsenical dimethylarsinic acid (DMA^V). The thio-metabolites dimethylmonothioarsinic acid (DMMTA^V) and monomethylmonothioarsinic acid (MMMTA^V) were also found in the urine of As^III-treated rats. The LC₅₀ concentrations of DMMTA^V for rat and human urothelial cells in vitro were similar to trivalent oxygen-containing arsenicals. These data suggest that dietary As^III-induced urothelial cytotoxicity and proliferation are dose responsive, and the urothelial effects have a threshold corresponding to the urinary excretion of measurable reactive metabolites.