Massive hemorrhage from the alimentary tract in a cirrhotic patient presents a difficult diagnostic and therapeutic problem. When the bleeding is due to varices they are usually either gastric or esophageal in location. We have recently had the opportunity to study a patient known to have esophageal varices for at least three years, in whom a duodenal varix, well demonstrated by barium meal examination and splenoportography, was largely responsible for the hemorrhage that led to death. Case Report This 65-year-old white man was first admitted to Strong Memorial Hospital in June 1959, after three days of black tarry stools. A long history of alcohol consumption was elicited. Physical examination demonstrated hepatosplenomegaly. Stool and gastric aspirant were strongly guaiac-positive. Hematocrit at admission was 30. The patient continued to pass blood per rectum and required transfusion with 6 units of whole blood over the next few days. Liver function tests demonstrated hepatocellular dysfunction. An upper gastrointestinal series revealed esophageal varices (Fig. 1A). The stomach, duodenal bulb, and duodenal loop (Fig. 2A) appeared normal. Esophagoscopy failed to confirm the presence of varices. After the insertion of a Sangstacken tube, however, the bleeding stopped and the patient's hematocrit became stabilized. He was discharged after seventeen days of hospitalization. For the next three years, the patient was apparently well although he drank about one pint of wine each day. Four days prior to his second and final admission to the hospital he again bled per rectum. On admission the hematocrit was 36. Laboratory evidence of hepatocellular damage was again found. Gastric aspiration was productive of faintly guaiacpositive material. An upper gastrointestinal series demonstrated esophageal varices (Fig. 1B). The stomach and duodenal bulb were normal. A lobulated defect 2 cm. in diameter was seen in the second portion of the duodenum (Fig. 2B). Later in the same day a splenoportogram was obtained. The preliminary radiograph of the abdomen demonstrated residual barium in the fundus of the stomach and small bowel. Contrast medium in the second portion of the duodenum again outlined the lobulated mass (Fig. 3A). After the injection of 30 c.c. of 50 per cent Hypaque into the splenic pulp both the splenic and portal veins filled. The peripheral radicals of the portal vein had an attenuated appearance. The coronary vein filled. In addition to this, a large vessel, which appeared to be the superior mesenteric vein, was opacified and shown to fill the defect in the duodenal loop (Fig. 3B). Leading from this defect, which was apparently a large varix, another vein coursed to the region of the liver bed. On the basis of these examinations a diagnosis of a varix of the second portion of the loop was made. During the next three days the patient received 22 units of blood.