Drug-induced liver injury (DILI) is an increasing health problem and a challenge for physicians, regulatory bodies and the pharmaceutical industry, not only because of its potential severity and elusive pathogenesis but also because it is often inaccurately diagnosed, commonly missed entirely and more often not reported. The general view is that idiosyncratic DILI, which is not predictable whether based on the pharmacology of the drug or on the dose administered, is determined by the presence in the recipient of variants in, or expression of, genes coding for key metabolic pathways and/or the immune response, and the interaction of these genetic variants with environmental variables. Furthermore, idiosyncratic DILI is an example of a complex-trait disease with two or more susceptibility loci, as reflected by the frequency of genetic variants in the population often being higher than the occurrence of significant liver injury. Polymorphisms of bioactivation/toxification pathways via the CYP450 enzymes (Phase I), detoxification reactions (Phase II) and excretion/transport (Phase III), together with immunological factors that might determine DILI are reviewed. Challenges such as gene-trait association studies and whole-genome studies, and future approaches to the study of DILI are explored. Better knowledge of the candidate genes involved could provide further insight for the prospective identification of susceptible patients at risk of developing drug-induced hepatotoxicity, development of new diagnostic tools and new treatment strategies with safer drugs.