As part of a large multicenter Veterans Administration Cooperative Study of Alcoholic Hepatitis, 89 patients with clinically mild biopsy-proven disease were followed for at least 30 months. Although clinical and laboratory abnormalities were minimal, cirrhosis was present in 38%, and mortality was 22% at 30 months. Clinical features suggesting more advanced disease (i.e., ascites and encephalopathy) and laboratory parameters for the diagnosis of alcoholic hepatitis and/or cirrhosis were imprecise and frequently misleading. The histologic diagnosis of cirrhosis correlated best with changes in immunoglobulin A, prothrombin time, and SGOT/SGPT. However, by using logistic discriminant analysis on 26 commonly available laboratory tests to diagnose cirrhosis, only a 72% sensitivity and 88% specificity could be obtained. Mortality in the patients with cirrhosis (10/34) was significantly higher at 1 and 2 yr compared with patients without cirrhosis (10/55, p less than 0.01). The high mortality in noncirrhotics may have resulted from progression to cirrhosis subsequent to the initial evaluation. Thus, liver biopsy in this population with minimal disease seems necessary to establish both an accurate diagnosis and the reversibility of the disease.