Drug-induced liver injury is a frequent cause of hepatic dysfunction. Reliably establishing whether the liver disease was caused by a drug requires the exclusion of other plausible causes and the search for a clinical drug signature. The drug signature consists of the pattern of liver test abnormality, the duration of latency to symptomatic presentation, the presence or absence of immune-mediated hypersensitivity and the response to drug withdrawal. Determination of causality also includes an evaluation of individual susceptibility to drug-induced liver injury. This susceptibility is governed by both genetic and environmental factors. Components of the drug signature in conjunction with certain risk factors have been incorporated into formal scoring systems that are predictive of the likelihood of drug-induced liver injury. The most validated scoring system is the Roussel-Uclaf causality assessment method, which nonetheless retains certain imperfections. Mitigating the potential for drug-induced liver injury is achieved by the identification of toxicity signals during clinical trials and the monitoring of liver tests in clinical practice. There are three signals of liver toxicity in clinical trials: (i) a statistically significant doubling (or more) in the incidence of serum alanine aminotransferase (ALT) elevation >3 x the upper limit of normal (ULN); (ii) any incidence of serum ALT elevation >8-10 x ULN; and (iii) any incidence of serum ALT elevation >3 x ULN accompanied by a serum bilirubin elevation >2 x ULN. Monitoring of liver tests in clinical practice has shown unconvincing efficacy, but where a benefit-risk analysis would favour continued therapy, monthly monitoring may have some benefit compared with no monitoring at all. With rare exception, treatment of drug-induced liver injury is principally supportive. Drug toxicity is the most common cause of acute liver failure, defined as a prolonged prothrombin time (international normalised ratio > or =1.5) and any degree of mental alteration occurring <26 weeks after the onset of illness in a patient without pre-existing cirrhosis. A patient who meets these criteria must be evaluated for liver transplantation. The pathogenesis of drug-induced liver injury can be examined on the basis of the two principal patterns of injury. The hepatocellular pattern is characterised by a predominant rise in the level of transaminases and results from the demise of hepatocytes by means of either apoptosis or necrosis. The cholestatic pattern is characterised by a predominant rise of the serum alkaline phosphatase level and usually results from injury to the bile ductular cells either directly by the drug or its metabolite, or indirectly by an adaptive immune response.