Drug hepatotoxicities have been evaluated in case histories, surveys based on retrospective record reviews, and spontaneous adverse drug reactions reported to national pharmacovigilance systems, but in relatively few epidemiologic studies. To identify and quantify the risk of acute liver injury associated with individual drugs, we reviewed and integrated all the published epidemiologic research on the subject. The source population for the eight studies was general population registered on a single large general practice-based computerized data base. All were retrospective cohort studies, but some had a case-control design nested within the source cohort. Participants were selected according to their use of selected agents (nonsteroidal antiinflammatory drugs [NSAIDs], antibiotics, acid-suppressing drugs, other drugs suspected of being hepatotoxic) during the study period. Among the agents, we found a group of important hepatotoxic drugs with an associated incidence rate of acute liver injury greater than 100/100,000 users, including chlorpromazine and isoniazid. Agents with less risk but greater than 10/100,000 users were amoxicillin-clavulanic acid and cimetidine. A third group of drugs had an associated incidence rate of acute liver injury of less than 10/100,000 users. Our results provide evidence of relative safety for commonly administered agents such as NSAIDs, amoxicillin, omeprazole, and ranitidine. We also quantified important suspected liver toxicity, providing a reasonable precise risk estimate of clinical liver injury associated with chlorpromazine, isoniazid, and amoxicillin-clavulanic acid.