-Blockers and sclerotherapy prevent long-term upper digestive rebleeding in cirrhosis but they seem ineffective for early rebleeding. We compared octreotide with a placebo for the prevention of early rebleeding in cirrhotic patients. After control of acute upper digestive bleeding, 262 consecutive cirrhotic patients were randomized to octreotide 100 g subcutaneously three times a day for 15 days (n ؍ 131) or to the placebo (n ؍ 131), in a double blind pragmatic trial in which -blockers and/or sclerotherapy were allowed together with the experimental treatment. Separate randomization and analysis were performed according to whether patients were eligible for -blockers and/or sclerotherapy (101 placebo, 97 octreotide) or not (30 placebo, 34 octreotide). Rebleeding within 15 days was the primary measure of treatment efficacy; 6-week rebleeding rate was also assessed as a secondary measure. Fifteen-day cumulative proportions of patients rebleeding were 28% in the placebo group and 24% in the octreotide group (P ؍ .40); corresponding figures among the 198 patients eligible to -blockers and/or sclerotherapy were 26% and 16% (P ؍ .05) and among the 64 not eligible for these treatments 33% and 49% (P ؍ .29). Among patients eligible to -blockers and/or sclerotherapy, a significant reduction of rebleeding episodes (35 vs. 18, P ؍ .03), blood transfusions (75 vs. 50, P ؍ .04), and days of stay in hospital (1,544 vs. 1,190, P ؍ .0001) was also found in the octreotide group: this beneficial effect was confirmed 6 weeks after randomization. Following upper digestive bleeding in cirrhosis, almost 50% of patients rebleed and 30% die within 6 weeks.