Although several mouse models of AIH have been described, no model is ideal. Indeed, the disease is self-limited in each model, and none is associated with significant liver fibrosis or progression to cirrhosis. Nevertheless, these models should be useful for testing different hypotheses regarding the initiation of AIH. Still, each model poses unique limitations. The EAIH model initiated by immunization with crude liver antigens in CFA has been plagued by a high prevalence of hepatitis lesions in CFA controls and inconsistencies in results. The TGF beta-1 and IL-2 deficient models lead to high in utero mortality and short median life spans in the surviving animals. Lastly, all models require more detailed characterization of the antigen specificity of infiltrating liver T cells and the pathogenesis of liver cell injury.