Propylthiouracil has been shown experimentally to protect against alcohol-induced hepatocellular necrosis in hypoxic conditions. An earlier, short-term study of patients with alcoholism and liver disease indicated clinical improvement with propylthiouracil, but the effect on mortality could not be assessed. In the present study, we investigated the effect of propylthiouracil on mortality in patients with alcoholic liver disease in a long-term, double-blind, randomized clinical trial involving 310 compliant patients who received propylthiouracil (n = 157) or placebo (n = 153) for a maximum of two years. There were no differences between the two groups in demographic and clinical characteristics and biopsy-confirmed diagnoses at randomization, or in daily urinary alcohol levels during the study. The cumulative dropout rate over two years was not significantly different (propylthiouracil group, 0.68; placebo group, 0.60). The group receiving propylthiouracil (300 mg per day) had a cumulative mortality rate half that in the group receiving placebo (0.13 vs. 0.25 [P less than 0.05] in the total sample, and 0.25 vs. 0.55 [P less than 0.03] in a subgroup of severely ill patients [propylthiouracil group, n = 56; placebo group, n = 41]). Proportional-hazards stepwise regression analyses indicated that only propylthiouracil treatment, prothrombin time, hemoglobin levels, and mean daily urinary alcohol levels significantly affected mortality. The hazards ratio for the complete group indicated that mortality in the propylthiouracil group was 0.38 (95 percent confidence interval, 0.20 to 0.83) that of the placebo group. Protection by propylthiouracil was not observed in patients with high morning urinary alcohol levels. No clinically important side effects of propylthiouracil were observed at the dose used. We conclude that the administration of propylthiouracil can reduce mortality due to alcoholic liver disease.