OBJECTIVE An earlier pilot study from our liver unit suggested benefit from treatment with pentoxifylline (PTX), an inhibitor of tumor necrosis factor (TNF), in severe acute alcoholic hepatitis. The aim of the present study was to evaluate this treatment in a larger cohort of patients. METHODS One hundred one patients with severe alcoholic hepatitis (Maddrey discriminant factor > or = 32) entered a 4-week double-blind randomized trial of PTX (400 mg orally 3 times daily) vs. placebo. Primary endpoints of the study were the effect of PTX on (1) short-term survival and (2) progression to hepatorenal syndrome. On randomization, there were no differences in demographic and clinical characteristics or laboratory values (including TNF) between the 2 groups. RESULTS Twelve (24.5%) of the 49 patients who received PTX and 24 (46.1%) of the 52 patients who received placebo died during the index hospitalization (P = 0.037; relative risk, 0.59; 95% confidence interval, 0.35-0.97). Hepatorenal syndrome was the cause of death in 6 (50%) and 22 (91.7%) patients (P = 0.009; relative risk, 0.29; 95% confidence interval, 0.13-0.65). Three variables (age, creatinine level on randomization, and treatment with PTX) were independently associated with survival. TNF values on randomization were not predictive of survival; however, during the study period they increased markedly in nonsurvivors compared with survivors in both groups. CONCLUSIONS Treatment with PTX improves short-term survival in patients with severe alcoholic hepatitis. The benefit appears to be related to a significant decrease in the risk of developing hepatorenal syndrome. Increasing TNF levels during the hospital course are associated with an increase in mortality rate.