There is substantive evidence that most idiosyncratic drug reactions (IDRs) are immune mediated. That raises the question: by what mechanism do drugs induce an immune response. Although it is difficult to prove, there is evidence that many IDRs are caused by chemically reactive drugs or reactive metabolites of drugs. For example, ß-lactams covalently bind to proteins, and this can lead to the production of IgE antibodies that recognize drug-modified proteins and can cause an anaphylactic reaction. The covalent binding of a drug or reactive metabolite to proteins produces neoantigens that can be recognized as foreign. This is the basis of the hapten hypothesis. However, foreign proteins do not generally induce a strong immune response without an adjuvant to activate antigen presenting cells (APCs). This activation upregulates costimulatory molecules on APCs that are required to activate naive T cells that recognize the drug-modified proteins. A complementary hypothesis to the hapten hypothesis is the danger hypothesis, which states that unless something causes damage to an organism the immune system will ignore it. Damaged cells can release danger associated molecular pattern molecules (DAMPs) that activate APCs. These are the two basic hypotheses of IDRs; however, they remain to be rigorously tested, and immune responses are much more complex. It is virtually impossible to do controlled experiments in humans to test IDR hypotheses. Animal models are essential for controlled experiments, but IDRs are also idiosyncratic in animals, and therefore there has been a lack of valid animal models. One exception is nevirapine-induced skin rash in brown Norway rats. It appears that the dominant immune response to drugs that can cause IDRs is immune tolerance, especially in the liver. Therefore we blocked the checkpoints PD-1 and CTLA-4 to impair immune tolerance and found that this produced an animal model of amodiaquine-induced liver injury similar to the injury that occurs in humans. In particular there is a delay in onset of about 3 weeks, and the histology is similar to that of idiosyncratic liver injury in humans with piecemeal necrosis and an inflammatory infiltrate containing CD8 T cells. The injury was prevented by depletion of CD8 T cells, which indicates their involvement in the injury. Impairment of immune tolerance unmasked the ability of other drugs to cause liver injury such as nevirapine, isoniazid, troglitazone, and tolcapone, although the injury was milder than that caused by amodiaquine. We also found that the supernatant from the incubation of drugs with hepatocytes activated inflammasomes in macrophages, which is consistent with the danger hypothesis, but we do not know the identity of the presumed DAMPs in the supernatant. These studies provide an animal model that can be used to test mechanistic hypotheses such as the involvement of mitochondria in the production of DAMPs. In addition, it opens the possibility that the animal model or even the in vitro assay would be able to predict which drug candidates are likely to cause IDRs based on their mechanisms.