The nonsteroidal antiinflammatory drug diclofenac causes rare but significant cases of serious hepatotoxicity, typically with a delayed onset (>1-3 months). Because there is no simple dose relationship and because liver injury cannot be reproduced in current animal models, individual patient-specific susceptibility factors have been evoked to account for the increased risk. While these patient factors have remained undefined, a number of molecular hazards have been characterized. Among these are metabolic factors (bioactivation by hCYP2C9 or hCYP3A4 to thiol-reactive quinone imines, activation by hUGT2B7 to protein-reactive acyl glucuronides and iso-glucuronides, and 4'-hydroxylation secondary to diclofenac glucuronidation), as well as kinetic factors (Mrp2-mediated concentrative transport of diclofenac metabolites into bile). From the toxicodynamic view, both oxidative stress (caused by putative diclofenac cation radicals or nitroxide and quinone imine-related redox cycling) and mitochondrial injury (protonophoretic activity and opening of the permeability transition pore) alone or in combination have been implicated in diclofenac toxicity. In some cases, immune-mediated liver injury is involved, inferred from inadvertent rechallenge data and from a number of experiments demonstrating T cell sensitization. Why certain underlying diseases (e.g., osteoarthritis) also increase the susceptibility to diclofenac hepatotoxicity is not clear. To date, cumulative damage to mitochondrial targets seems a plausible putative mechanism to explain the delayed onset of liver failure, perhaps even superimposed on an underlying silent mitochondrial abnormality. Increased efforts to identify both patient-specific risk factors and disease-related factors will help to define patient subsets at risk as well as increase the predictability of unexpected hepatotoxicity in drug development.