A retrospective analysis of results obtained over a three year period in 66 patients receiving selective intra-arterial vasopressin (SIAV) for control of 69 episodes of massive gastrointestinal bleeding was presented. SIAV was used when there was a failure of conventional medical therapy and the patient's pathology and/or complicating medical conditions necessitated an attempt at controlling an emergent condition by nonoperative means. Hemmorrhage was completely controlled in 43% of variceal bleeds, 67% of hemorrhage gastritis, 45% of bleeding ulcers, and in 62% of colonic sources. The incidence of rebleeding following initial control was 16%. The surgical mortality for patients who were initial failures of SIAV was 50%. Patients undergoing elective surgery after complete control by SIAV had an 8% mortality. There were five catheter related complications. Minor complications occurred in 41% of patients, but required no treatment. Major complications occurred in 40% of cirrhotic and 21% of noncirrhotic bleeding episodes; and were a contributing factor in five cirrhotic deaths and three noncirrhotic deaths. In critically ill patients in the setting of an Intensive Care Unit, selective intra-arterial vasopressin appears: 1) to be an effective means of controlling certain types of gastrointestinal hemorrhage; 2) to provide an opportunity for an increase in survival rate.