Background & AimsPentoxifylline, an inhibitor of tumor necrosis factor-α, is given to patients with liver diseases, but its effects in patients with advanced cirrhosis are unknown. We performed a randomized, placebo-controlled, double-blind trial of its effects in patients with cirrhosis. Methods A total of 335 patients with cirrhosis (Child–Pugh class C) were assigned to groups given either pentoxifylline (400 mg, orally, 3 times daily; n = 164) or placebo (n = 171) for 6 months. The primary end point was mortality at 2 months. Secondary end points were mortality at 6 months and development of liver-related complications. Results By 2 months, 28 patients in the pentoxifylline group (16.5%) and 31 in the placebo group (18.2%) had died (P = .84). At 6 months, 50 patients in the pentoxifylline group (30.0%) and 54 in the placebo group (31.5%) had died (P = .75). The proportions of patients without complications (eg, bacterial infection, renal insufficiency, hepatic encephalopathy, or gastrointestinal hemorrhage) were higher in the pentoxifylline group than in the placebo group at 2 months (78.6% vs 63.4%; P = .006) and 6 months (66.8% vs 49.7%; P = .002). The probability of survival without complications was higher in the pentoxifylline group than in the placebo group at 2 and 6 months (P = .04). In multivariate analysis, the factors associated with death were age, the Model for End-Stage Liver Disease score, and presence of early-stage carcinoma. Treatment with pentoxifylline was the only factor associated with liver-related complications. Conclusions Although pentoxifylline does not decrease short-term mortality in patients with advanced cirrhosis, it does reduce the risk of complications.