This (1)H nuclear magnetic resonance metabonomics study was aimed to determine urinary biomarkers of cholestasis resulting from inhibition of biliary secretion of bile or obstruction of bile flow. To inhibit biliary secretion of bile, cyclosporine A was administered to male Sprague-Dawley rats. Obstruction of bile flow was induced by administration of 4,4'-methylene dianiline, alpha-naphthylisothiocyanate or bile duct ligation. Clinical pathological and histopathological examinations were performed to confirm cholestatic injury and (1)H nuclear magnetic resonance spectral data for urine samples were analysed to determine similarities and differences in profiles of metabolites using the Spotfire. In cyclosporine A-treated groups, serum total bilirubin and bile acid were significantly increased but no remarkable hepatic histopathological-changes were observed. In 4,4'-methylene dianiline-, alpha-naphthylisothiocyanate- and bile duct ligation-treated groups, serum alkaline phosphatase, gamma-glutamyltranspeptidase and total bilirubin levels increased significantly, and hepatic histopathological-changes were observed. On urinary (1)H nuclear magnetic resonance spectral analysis, area intensities derived from 0.66 to 1.90 ppm were decreased by cyclosporine A, whereas they were increased by other treatments. These metabolites were identified using the NMR suite as bile acids, branched-chain amino acids, n-butyrate, propionate, methyl malonate and valerate. These metabolites were further investigated by K-means clustering analysis. The cluster of these metabolites is considered to be altered by cholestasis. We conclude that bile acids, valine and methyl malonate have a possibility to be urinary cholestatic biomarkers, which distinguish a difference in mechanism of toxicity. (1)H nuclear magnetic resonance metabonomics thus appears to be useful for determining the mechanisms of toxicity and can be front-loaded in drug safety evaluation and biomarker discovery.