The infrequent occurrence of idiosyncratic reactions and their dependence on individual sensitivity factors allow them to go undetected in current preclinical safety evaluation using conventional animal tests. Better predictive models for idiosyncratic, drug-induced liver injury (IDILI) would enable the preclinical elimination of drug candidates with idiosyncrasy liability and could provide evidence for a mode of action for these responses, suggest early biomarkers of IDILI, and lead to the development of mechanism-based, in vitro screens. Desirable characteristics of an animal model include the production of liver injury in a large fraction of animals of relatively inexpensive species/strains and the ability to distinguish drugs that cause IDILI in humans from ones that do not. The mechanistic basis for idiosyncratic reactions remains poorly understood. However, attempts at animal model development have been made based on several hypothesized modes of action of IDILI. These hypotheses have centered on drug disposition polymorphisms, adaptive immunity, mitochondrial dysfunction, failure to adapt to modest injury, inflammatory stress, and multiple determinants, and the success in achieving animal models of liver injury for each of these is discussed. Despite numerous challenges associated with animal models of IDILI, some models have emerged and are proving useful in exploring potential mechanisms. Current animal models are not perfect, but they hold promise for increasing the prediction and understanding of human idiosyncratic drug reactions.