To assess if propranolol prevents the first bleeding in cirrhosis, we randomly assigned 174 patients with large varices to either propranolol in doses reducing the resting heart rate by 25% (85 patients) or to a placebo (oral vitamin K, 89 patients). Sixty-nine patients had alcoholic cirrhosis, 24 posthepatitic cirrhosis and 81 cryp-togenic cirrhosis. At the time of inclusion, 75 patients (43%) had ascites and according to the Child-Pugh classification, 103 (59%) had Class A disease, 60 (34%) Class B and 11 (7%) Class C. We report here an interim analysis of the study when all patients had been followed for at least 1 year (mean follow-up = 22 months). At this time, the cumulative proportion of patients free of bleeding was 74% in the propranolol group and 63% in the placebo group; corresponding survival figures were 59 and 74% (both differences not significant), respectively. A retrospective analysis showed that the cumulative percentage of patients free of bleeding was significantly higher in the propranolol- than in the control-group in the subsets of patients without ascites or in Child-Pugh Class A: respectively, 87 vs. 64% (p = 0.023) and 88 vs. 64% (p = 0.01). No differences in bleeding incidence were found in patients with ascites or in Child-Pugh Class B or C. Propranolol treatment did not affect survival in any subgroup. Twenty-five patients had to be withdrawn from propranolol because of side effects (n = 23) or low compliance (n = 2). If confirmed on a longer follow-up, these results suggest that propranolol could prevent the first bleeding in patients with well-compensated cirrhosis.