Bile acids (BAs) are crucial for the diagnosis, follow-up, and prognostics of liver and intestinal disorders and other diseases affecting BA metabolism. A rapid, simple, and sensitive analytical method is needed to demonstrate the full metabolic profile and simultaneously determine the individual BAs in biological samples. In our present study, an ultraperformance liquid chromatography-mass spectrometry (UPLC-MS) method has been established and validated for simultaneous quantitation of 22 BAs and a metabonomic study was performed based on the chemometric analysis of the serum samples from carbon tetrachloride (CCl4)- and alpha-naphthylisothiocyanate (ANIT)-induced liver failure rats. The optimal chromatographic condition was effected by UPLC (Acquity UPLC BEH column, 1.7 microm, 2.1 mm x 100 mm) using a linear gradient elution system of methanol-5 mM ammonium acetate containing 0.01% acetic acid after a simple-step deproteinization by precipitation. The separation of the 22 BAs can be finished in less than 12 min, and the concentrations of these BAs in rat serums were simultaneously determined using a selective ions monitoring mode. The method was validated with respect to repeatability (relative standard deviation < 9.78%) and accuracy (relative errors from -13.55 to 9.58%). The range of each BA was found from not detected (nd) to 8301 ng mL(-1), respectively. Furthermore, the developed method was successfully applied to the metabonomics analysis of BAs in CCl4- and ANIT-induced liver failure rats, using principle component analysis and canonical discriminant analysis. The serum samples from the two types of rat liver injury could be distinguished from each other and from the untreated animals according to the varieties of BAs. It indicated that the level of BAs could be considered as a sensitive parameter of hepatotoxicity induced by different chemical toxins. This novel metabonomics study of BAs based on the UPLC-MS profile provides not only an accurate quantitative assay of the serum concentrations of biomarkers but also a promising methodology for evaluation of liver injury.