We conducted a prospective, randomized, multicenter, single-blind trial of propranolol as compared with placebo in the prevention of first upper gastrointestinal tract bleeding in patients with cirrhosis of the liver. A total of 230 patients (90 percent with alcoholism and 46 percent with a Child-Pugh grade C classification) with large esophageal varices without previous bleeding were randomly assigned to receive either propranolol (n = 118) or placebo (n = 112), after they had been divided into two groups according to the severity of their liver disease. The end points of the study were bleeding and death. The dose of propranolol was progressively increased to decrease the heart rate by 20 to 25 percent. The final doses were 40 mg of conventional propranolol and 160 and 320 mg of long-acting propranolol daily in 22 percent, 60 percent, and 18 percent of patients, respectively. The mean (+/- SD) follow-up time among survivors without bleeding was 436 +/- 172 days. The cumulative percentages of patients free of bleeding two years after inclusion in the study were 74 percent (95 percent confidence limits, 61 and 83) in the propranolol group and 39 percent (95 percent confidence limits, 15 and 69) in the placebo group (P less than 0.05). Cumulative two-year survival was 72 percent (95 percent confidence limits, 60 and 81) in the propranolol group and 51 percent (95 percent confidence limits, 37 and 64) in the placebo group (P less than 0.05). The advantage of propranolol over placebo was maintained when potentially confounding variables were adjusted with use of the Cox model. Side effects occurred in 17 percent of the patients who received propranolol and led to the stopping of treatment in 11 percent. We conclude that propranolol can decrease the incidence of first bleeding and death during a period of two years in patients with cirrhosis and large varices.