To analyze the relationship between the splanchnic and systemic effects of vasopressin and to measure its efficacy in lowering portal pressure relative to what can be accomplished by zero gradient shunting, intraoperative measurements of cardiac output and relevant pressures were made in 30 patients undergoing selective or total shunts. Vasopressin caused a significant increase in systemic vascular resistance and pulmonary capillary wedge pressure, but an insignificant overall reduction in cardiac index (CI). However, in ten patients the decrease in CI exceeded 20%, suggesting a subpopulation of especially susceptible individuals. High initial CI, age, pre-existent heart disease, and severity of cirrhosis did not predict greater vulnerability. Adding an infusion of nitroprusside regularly reverted CI to control levels, regardless of the extent of cardiac output depression. Vasopressin was 38% as effective as a subsequent shunt in reducing splanchnic venous pressure. The portal hypotensive action bore no relationship to CI, but the pressure decrease caused by vasopressin was predictive of the reduction that could be achieved by shunting. The effects of the two types of shunts on systemic hemodynamics were minor and remarkably similar.