In a double blind randomized study, branched-chain amino acids and placebo (casein) were compared as a treatment for chronic hepatic encephalopathy in cirrhosis. After a 15-day run-in period with controlled diet (45-65 g protein), the patients were administered, in addition to their diet, branched-chain amino acids (0.24 g/kg, 30 patients) or an equinitrogenous amount of casein (34 patients). One patient on branched-chain amino acids and two on casein were lost to the study. After 3 months, the index of portal-systemic encephalopathy significantly improved in patients on active treatment (from 40 [S.D. 14]% to 21 [17]), but was not in subjects receiving casein (from 37 [13]% to 36 [12]). Two or more parameters of the index improved in 24 patients treated with amino acids (80%; confidence limits, 61-92%), and only in 12 receiving casein (35%; confidence limits, 20-54%; p less than 0.001). Patients who did not improve were given an alternative treatment for 3 more months. Casein-treated patients given branched-chain amino acids rapidly improved. The changes in neuropsychologic function were associated with an improvement in semiquantitative nitrogen balance, which became consistently positive in amino acid-treated subjects; there was also a mild improvement in nutritional parameters and in liver function tests. The supplementation of oral branched-chain amino acids to the diet is superior to casein as a treatment for providing adequate nitrogen supply and improving the mental state of cirrhotic patients with chronic encephalopathy.