We conducted a prospective randomized controlled study to evaluate the effectiveness of propranolol and sclerotherapy, compared to a control group, in the prevention of variceal rebleeding in alcoholic cirrhotic patients. Among the 79 patients included, the distribution of patients according to Child-Pugh classification was: A, 22%; B, 40%; and C, 38%. Propranolol was given twice daily with a mean final dose of 54 +/- 16 mg/day, this resulted in a mean reduction in resting heart rate of 26 +/- 7%. Sclerotherapy was performed weekly using 1% polidocanol. End points were rebleeding or death. During the mean follow up of 19 +/- 16 months, 43 patients bled and 22 patients died. The cumulative percentages of patients free of rebleeding at 1 year were: propranolol, 81% (95% confidence interval (CI): 63-92); sclerotherapy, 64% (95% CI: 45-82); control, 54% (95% CI: 36-71); these differences did not reach statistical significance. The cumulative percentages of patients alive at 1 year were: propranolol, 92% (95% CI: 76-98); sclerotherapy, 79% (95% CI: 58-91); control, 81% (95% CI: 60-93); these differences were not statistically significant. Alcohol withdrawal, which occurred in 66% of patients, was an independent predictive factor associated with a decreased risk of rebleeding or death. In conclusion, a life table analysis of patients free of rebleeding, as well as of patients surviving, revealed a tendency in favour of propranolol. The lack of a statistical support for these two favorable effects could be due to poor statistical power.(ABSTRACT TRUNCATED AT 250 WORDS)