74 cirrhotic patients with a history of variceal or gastric bleeding were randomly assigned to treatment with propranolol (40 to 360 mg/day) or placebo. The patients were all in good condition and doses of propranolol were titrated until a 25% reduction in heart rate was achieved. After 2 years, the cumulative percentage of patients free from rebleeding was significantly greater among the patients receiving propranolol (79%) than in the placebo group (32%; p less than 0.0001). Similarly, the percentage of surviving patients was significantly greater with propranolol (90%) than with placebo (57%; p less than 0.02) after 2 years. It was concluded that in cirrhotic patients in good condition, propranolol reduced both the risk of recurrent gastrointestinal haemorrhage and the mortality rate during a 2-year period of continuous administration of the drug.