A controlled randomized study was performed in 15 patients with biopsy-proven alcoholic hepatitis to determine the effect of the administration of a parenteral amino acid-glucose solution for 1 month on nutritional, clinical, biochemical and histological parameters. All patients were allowed ad libitum consumption of a hospital diet. Five patients received the amino acid-glucose solution, while 10 received the glucose solution without amino acids. There was more improvement in nitrogen balance in the treated group than the control group. However, amino acid therapy was no more beneficial than control therapy in improving creatinine-height index, arm muscle area, arm fat area, and plasma levels of retinol binding proteins, prealbumin and pyridoxal-5'-phosphate. Clinical and biochemical markers of liver disease improved in both groups. This improvement in composite clinical index was more rapid in the treated than in the control group, but this early advantage was no longer apparent at the end of one month. Hepatocellular necrosis, inflammation and fat improved in the entire group. Amino acid treatment resulted in a greater resolution of fatty infiltration, but did not otherwise affect hepatic histology. Hepatocellular necrosis and inflammation in both the initial and final liver biopsies were highly correlated with nitrogen balance on admission. Initial clinical index correlated with the quantity of ethanol consumption prior to admission but not with liver injury on biopsy. However, by the end of the study, clinical index correlated with hepatocellular necrosis and inflammation.(ABSTRACT TRUNCATED AT 250 WORDS)