To determine the feasibility of the use of Medicaid data to study drug-induced acute liver disease, we reviewed the medical records of 414 patients receiving Medicaid, age 20 or older, with an ICD-9-CM inpatient billing code consistent with acute hepatitis. Of the patients whose records were reviewed, 15.9% were alcoholics, 31.9% had acute hepatitis A or B, 13.5% were intravenous drug users, 8.2% had acute cholecystitis or choledocholithiasis, and 4.1% had received a blood transfusion within the previous 6 months. No diagnosis of liver disease was found in 10.6% of the patients, and 5.7% had chronic liver disease. Of the 169 patients with idiopathic acute liver disease identified, many had very mild liver disease and were hospitalized for reasons other than liver disease. We conclude that Medicaid billing data has high reliability and validity for the diagnosis of acute liver disease. However, primary medical records are essential for the study of drug-induced hepatitis, to be able to exclude other causes of liver disease, and to obtain information not included in the computer data.