Ketoconazole has only recently been recognized as a cause of hepatic injury, with most reports coming from outside the United States. In order to characterize more fully the U.S. experience, we undertook an analysis of 54 reports of alleged ketoconazole-induced liver injury submitted to the Food and Drug Administration from the time of initial marketing in 1980. Thirty-three reports were considered likely instances of ketoconazole-induced hepatitis. The majority of these cases occurred in women more than 40 yr of age. Jaundice was recorded in 27 individuals after therapy of 11-168 days with an average daily dose of 200 mg. Anorexia, malaise, nausea, and vomiting accompanied liver injury in one-third of cases. No instances of rash or eosinophilia were recorded. Serum transaminase and alkaline phosphatase values were consistent with acute hepatocellular injury in 18 patients, with primarily cholestatic injury in 5 patients, and with a mixed pattern in 9 individuals. Only one death seemed attributable to ketoconazole. In that patient, the drug was continued after the appearance of clinical and biochemical evidence of hepatic injury and massive hepatocellular necrosis was present at autopsy. The incidence of symptomatic, potentially serious hepatic injury appears to be very low, perhaps 1 in 15,000 exposed individuals. The presumed mechanism of injury is metabolic idiosyncrasy, although hypersensitivity has not been completely dismissed in some cases reported in the literature. The incidence of mild, asymptomatic, reversible elevations in serum transaminases occurring in ketoconazole recipients has been estimated to be 5%-10%. Periodic biochemical testing and monitoring for symptoms of hepatitis during ketoconazole therapy is recommended to help prevent the development of serious or fatal hepatic injury.