In a randomized, controlled trial to investigate the possible benefit of insulin and glucagon therapy in severe acute alcoholic hepatitis, 86 patients were randomized to receive 30 U insulin and 3 mg glucagon in 250 ml 5% dextrose over 12 hr each day for 3 wk or a similar regime of identical placebo. No significant differences were seen in patients' clinical characteristics and disease severity in the treated and placebo groups. Of the 43 patients receiving insulin and glucagon, 15 (35%) died within 4 wk of randomization, compared with 14 deaths (33%) in the control patients (p = not significant). When the patients surviving the first 4 wk were examined there were five more deaths in the treatment group, compared with one death in the control group at 6-mo follow-up (p = not significant). No significant differences in the frequency of short-term or long-term complications of alcoholic liver disease or relapse to alcohol were seen when the two groups were compared, although hypoglycemia was seen in six patients during infusion of insulin and glucagon. Similarly, no significant differences were seen in the improvement in clinical or biochemical features at 4 wk and at 6 mo in survivors when the insulin and glucagon-treated patients were compared with patients in the placebo group. This study does not confirm previous reports that insulin and glucagon infusion improves the outcome of severe acute alcoholic hepatitis.