Autoimmune hepatitis (AIH) is a complex liver disease with an increasing prevalence in recent years and can develop into the severe or fulminant form if the patients are not diagnosed accurately or treated in time. However, AIH accurate diagnosis, especially at the early stage, is still difficult to perform due to the absence of specific diagnostic markers and the large heterogeneity of its clinical, laboratory and histological features. To evaluate the biochemical process of AIH at the early stage, we investigated serum metabolic alterations in mice with liver injury induced by concanavalin A (Con A), which closely mimics the immune and inflammatory response of AIH in humans. Metabonomic profiling was performed by ultra-performance liquid chromatography quadrupole time-of-flight mass spectrometry (UPLC Q-TOF MS). As a result, fourteen metabolites were detected as potential biomarkers related to the early stage of the liver injury, including two bile acids (taurocholic acid (V1) and taurochenodeoxycholic acid (V2)), three long-chain acylcarnitines (tetradecanoylcarnitine (V4), linoleyl carnitine (V8) and L-palmitoylcarnitine (V9)), seven glycerophospholipids (lysoPE (18 : 0/0 : 0) (V3), lysoPC (16 : 0) (V5), lysoPC (18 : 1) (V7), lysoPC (18 : 0) (V10), lysoPC (20 : 1) (V11), lysoPE (22 : 0/0 : 0) (V12) and lysoPC (20 : 0) (V13)), a bilirubin (V14), and a retinyl ester (V6). Moreover, partial least square regression analysis (RLS-RA) showed that metabolism of glycerophospholipids (P2), bile acids (P4) and retinol (P5) was highly correlated with the clinical outcomes, suggesting they played key roles in the early stage of the liver injury. Our results also demonstrated that a metabonomic approach coupled with PLS-RA is a powerful tool with which changes can be characterized in the levels of endogenous metabolites associated with disease progression and to assist in further understanding the molecular mechanism of the disease.