We performed a controlled trial of peripheral hyperalimentation in moderate and severe alcoholic hepatitis to determine whether improvement in survival and liver function could be obtained. Twelve patients with moderate and 22 with severe alcoholic hepatitis were randomized to 28 days of peripheral parenteral nutrition (PPN) or standard therapy (ST). In the moderate group, six were treated with each therapy. In the severe group, 10 were treated with PPN and 12 with ST. Routine liver tests, hepatocyte function (galactose elimination capacity), estimated hepatic blood flow (galactose clearance) and assessment of ascites and encephalopathy were performed at randomization and at 28 days. Groups were equally matched at randomization. In the moderate group PPN produced no improvement in morbidity (liver tests) and mortality (no deaths). In the severe group there were seven deaths (4 PPN, 3 ST). PPN produced greater improvement than ST in serum bilirubin and transferrin concentrations and a trend toward greater improvement in prothrombin time, serum albumin and galactose elimination capacity. PPN had no deleterious effect on encephalopathy or ascites as only ST patients developed ascites or encephalopathy after randomization. We conclude that PPN compared to ST (1) provides no benefit in moderate alcoholic hepatitis, but (2) did more rapidly improve morbidity (liver tests) and probably liver function in severe alcoholic hepatitis; (3) PPN did not improve early mortality, and (4) it had no deleterious effect on encephalopathy or ascites.