Metabolomics allows high-throughput analysis of low-molecular-weight compounds in biofluids that reflect the physiological status and biochemical metabolism of living systems. Hence it has the potential to evaluate toxicity and clarifies the metabolism-related toxic mechanisms. In this study a promising candidate drug parent, triptolide, was given to Sprague-Dawley rats as a model toxicant at a single dose of 0.6, or 2.4 mg/kg, i.g. Both routine biochemical assays and histopathological inspection showed timedependent hepatic toxicity at the higher dose, but no obvious toxicity at the lower dose. Meanwhile, serum metabolome was profiled using the non-targeted metabolomic tool, gas chromatography time-of-flight mass spectrometry. Based on the acquired metabolomic data, mathematical models were calculated and the metabolic patterns of serum were evaluated using projection to latent structure-discriminant analysis. The relative distance of each treated group from the normal control was calculated to provide a measure of toxicity. Treatment with triptolide at either the higher or lower dose caused deviations in the metabolic pattern and resulted in perturbation of taurine, creatinine, free fatty acids, b-hydroxybutyrate, tricarboxylic acid cycle intermediates, and amino acids. This finding indicates the dysfunction of b-oxidation of free fatty acids and impairment of the mitochondria and confirms the hepatic toxicity of triptolide. The identified toxic markers and the calculated relative distance values quantitatively demonstrated doseand time-dependent toxicity, whereas the scores plot of the model provided the qualitative information. The metabolomic approach was non-invasive and more sensitive than routine toxic assessment, and the results of both methods correlated well.