Seven patients with compensated liver cirrhosis and esophageal varices, all with a base line wedge hepatic vein pressure greater than 20 cm H2O, received 1-mg doses of vasopressin hormonogen (tGLVP) intravenously. There was a significant mean decrease in wedge pressure of 32%, which lasted for at least 20 min (the duration of measurement), with no change in cardiac output measured. The only cardiac response was a 10 to 20% bradycardia at the height of the moderate pressor response-otherwise the ECG was without change. In 5 patients who received the same tGLVP dose during surgery, direct measurements of portal venous pressure showed the same degree of decrease within 10 min of intravenous injection. Fifteen patients with liver cirrhosis and severe bleeding from esophageal varices were treated conservatively with blood transfusion and tGLVP as the only major drug aside from antibiotics. A nonrandomized control group of 13 patients with the same age distribution, stage of disease, number of previous bleeds, etc., was treated conservatively in the same manner, except that they received either no hemodynamically active drugs or short acting neurohypophysial peptide preparations such as Pitressin. In the control group there was a 61.5% total mortality, a 53.8% mortality directly related to uncontrollable bleeding, and a mean duration of the bleeding episode of 11 days. In the tGLVP-treated group total mortality was 20%, mortality directly related to uncontrollable bleeding was 13.3%, and mean duration of the bleeding episode was 2.9 days. These results appear to justify a large scale clinical trial of the vasopressin hormonogen in this disease.