A randomized, double-blind, placebo-controlled trial of somatostatin was conducted among 120 patients admitted for bleeding esophageal varices (59 placebo, 61 somatostatin). An initial 250-micrograms bolus of somatostatin followed by a 5-day continuous infusion of 250 micrograms/h and an identical administration of placebo were evaluated for both the control of bleeding and prevention of early rebleeding from varices. Failure to control bleeding occurred in 22 (36%) somatostatin patients vs. 35 (59%) placebo patients, with time to failure occurring earlier with placebo (P = 0.036). blood and plasma transfused per hour during drug infusion of trial drug was reduced in the somatostatin group: median 0.033 vs. 0.105 unit/h (P = 0.025). Use of balloon tamponade was halved in somatostatin-treated patients. The average effect of somatostatin was a 41% reduction in the hazard of failure (95% confidence interval, -1% to 65%, P = 0.0545) after adjustment for the severity of liver disease, which was the only other variable having a significant influence on time to failure. There was no difference in 30-day mortality per admission (7 placebo, 9 somatostatin) or complications. It is concluded that somatostatin is safe and more effective than placebo for the control of variceal bleeding.