In a prospective randomized study, 38 patients with massive upper gastrointestinal tract hemorrhage, mostly due to esophageal varices or erosive gastritis, were treated with either standard medical therapy or standard therapy plus selective intra-arterial vasopressin infusion. Cessation of hemorrhage occurred more frequently in the vasopressin-treated group. The study design did not permit meaningful comparisons of mortality of transfusion requirements. We conclude that in patients bleeding from esophageal varices or gastritis, selective intra-arterial vasopressin is more effective in controlling hemorrhage than standard therapy.