Idiosyncratic drug-induced liver injury (DILI) is a major clinical problem and poses a considerable challenge for drug development as an increasing number of successfully launched drugs or new potential drugs have been implicated in causing DILI in susceptible patient subsets. Although the incidence for a particular drug is very low (yet grossly underestimated), the outcome of DILI can be serious. Unfortunately, prediction has remained poor (both for patients at risk and for new chemical entities). The underlying mechanisms and the determinants of susceptibility have largely remained ill-defined. The aim of this review is to provide both clinical and experimental evidence for a major role of mitochondria both as a target of drugs causing idiosyncratic DILI and as mediators of delayed liver injury. We develop a unifying hypothesis that involves underlying genetic or acquired mitochondrial abnormalities as a major determinant of susceptibility for a number of drugs that target mitochondria and cause DILI. The mitochondrial hypothesis, implying gradually accumulating and initially silent mitochondrial injury in heteroplasmic cells which reaches a critical threshold and abruptly triggers liver injury, is consistent with the findings that typically idiosyncratic DILI is delayed (by weeks or months), that increasing age and female gender are risk factors and that these drugs are targeted to the liver and clearly exhibit a mitochondrial hazard in vitro and in vivo. New animal models (e.g., the Sod2(+/-) mouse) provide supporting evidence for this concept. However, genetic analyses of DILI patient samples are needed to ultimately provide the proof-of-concept.