Direct measurements of portal flow and pressure in a relatively large number of patients with cirrhosis show a marked reduction in flow associated with a nearly constant plateau of portal pressure. This lack of correlation indicates the complex relationships of resistances in the splanchnic, collateral, and hepatic circuits determining the division of the available splanchnic flow between the portal vein and the collateral pathways. Subtracting the measured portal flow from well-established estimates of total hepatic blood flow in cirrhosis suggests that the hepatic artery contributes more than one-half of the blood perfusing the cirrhotic liver. There was no instance of retrograde portal flow during the preshunt measurements, although such reversal was frequent after side-to-side portacaval anastomosis. Attempting to explain the plateau of portal pressure in the face of an increasing outflow resistance presumably associated with progress of the disease, we postulate that an augmented inflow resistance to the splanchnic chamber reduces splanchnic flow in cirrhosis. End-to-side portacaval anastomosis did not return normal portal flow, although it decreased pressure to accepted control levels. The assumption is that most of the splanchnic blood was flowing through the shunt, leading to a high splanchnic resistance in the immediate postshunt status. If this resistance was previously elevated, as suggested by the plateau of portal pressure, the mechanism responsible for the elevation was not immediately deactivated after the shunt, and the true effect of the operation upon splanchnic flow may not be measurable at such time. Respiratory oscillations were a significant component of portal flow in cirrhosis before and after portacaval anastomosis, indicating the limitations of any steady state analysis of the circulatory derangement in cirrhosis.