Propranolol has been reported to prevent the risk of hemorrhage in patients who survived episodes of variceal rupture. Since the first bleeding episode can be lethal, we did a prospective, randomized trial to see whether beta-blockers could also prevent the first hemorrhage. Seventy-nine consecutive cirrhotics with large esophageal varices by endoscopy and who had never bled were randomly allocated to one of the following treatments: placebo; ranitidine (300 mg per day), or nadolol (40 to 120 mg per day)--which is not cardio-selective, reduces portal hypertension and does not interfere with renal flow. Since no significant differences between ranitidine and placebo treatment were observed, the two groups were combined as the control group and compared with the nadolol group. After a mean follow-up of 24 months, only 1 of the 30 patients in the nadolol group had bled, while 11 of the 49 patients in the control group had bled. The percentages of patients who had not bled 1 and 2 years after the inclusion were 100 and 94.4% for the nadolol group and 81.2 and 70.2% for the control group (p less than 0.02), respectively. There were no differences in the mortality rate. In conclusion, nadolol significantly protects against the first gastrointestinal bleeding episode in cirrhotics.