Intravenous (IV) vasopressin has been used to control human upper gastrointestinal (GI) hemorrhage for over 30 years. Although the use of vasopressin has been studied extensively in adults, no study has evaluated its use in children. Vasopressin was used therapeutically in 15 episodes of esophageal variceal hemorrhage and two episodes of bleeding peptic ulcer. Nine of 17 episodes were controlled with vasopressin alone (53%). Balloon tamponade and variceal sclerosis were required for control in the remainder. Blood requirements averaged 53 mL/kg prior to control of hemorrhage. Metabolic complications occurred in 65% of the episodes. There were two groups of patients identified: those receiving greater or those receiving less than .01 units/kg/min of IV vasopressin. All of the complications identified occurred when greater than .01 U/kg/min of vasopressin were used (P less than .05). Control of bleeding was not improved with higher doses of vasopressin. These data suggest that the use of IV vasopressin at doses greater than .01 U/kg/min to control GI bleeding will increase the incidence of complications without improving control of hemorrhage.