To compare the efficacy of endoscopic paravariceal sclerotherapy and oral propranolol in the prevention of recurrent upper gastrointestinal bleeding, 78 cirrhotic patients were randomly assigned to either treatment after an endoscopically proven bleed from esophageal varices. After randomization, but before treatment had been started, a total of eight patients had to be withdrawn from the study due to early rebleeding (requiring emergency sclerotherapy) or violations of the protocol. Among the 70 patients analyzed (36 sclerotherapy, 34 propranolol), both treatment groups were comparable with respect to demographic, clinical and laboratory data. The groups also did not differ with respect to continued alcohol intake. Sclerotherapy was performed twice weekly using 1% polidocanol as the sclerosing agent until the varices were eradicated or well-covered by fibrous tissue. Propranolol was given twice daily at a dose reducing the resting heart rate by 25% (60 to 320 mg per day; mean ± SD = 161 ± 80 mg per day). Patients were followed for up to 2 years with visits at 3 monthly intervals (mean follow-up = sclerotherapy, 14 months; propranolol, 9.2 months). Life table analysis of patients without rebleeding from nonvariceal sites revealed a tendency in favor of propranolol; however, the difference did not reach statistical significance. No significant difference was observed between sclerotherapy and propranolol in the proportion of patients rebleeding from esophageal varices or from all sources of upper gastrointestinal bleeding. Furthermore, survival was similar in both treatment groups. A small esophageal perforation due to sclerotherapy was observed in one patient, and one propranolol patient had to discontinue the medication due to severe bradycardia. Other complications of sclerotherapy and side effects of propranolol were minor. Propranolol reduced variceal size significantly within 3 months of treatment. No such decrease was observed in a subgroup of patients who rebled under propranolol, and in the untreated control group of an ongoing trial of prophylactic sclerotherapy. We conclude that propranolol and sclerotherapy are of comparable value in preventing recurrent upper gastrointestinal bleeding in cirrhotics.