To establish the effects of different modes of cardiopulmonary bypass on hepatic blood flow, with the aim of increasing understanding of the hemodynamic factors that may lead to hepatic dysfunction in patients after cardiopulmonary bypass. The anatomical and physiologic characteristics that are unique to the hepatic circulation are also reviewed, together with an account of the known specific effects on the liver of several hemodynamic stimuli that are commonly present during cardiopulmonary bypass. The entire world literature on the subjects of clinical and experimental cardiopulmonary bypass, extracorporeal circulation, blood gas tensions, hypothermia, hypotension, and hepatic blood flow was searched via Index Medicus, up to and including 1991. Data are presented only from those studies that employed sound cardiopulmonary bypass or blood flow measurement technique. For investigations on hypothermia, data for temperatures < 28 degrees C are not included. Details of one recent experimental study in the dog by the author are highlighted, in which the relative effects of pump flow rate, temperature, and type of perfusion (pulsatile or nonpulsatile) are presented; the data are critically compared and contrasted with those data obtained by earlier workers. Hepatic hemodynamics are characterized by a dual supply of blood from the hepatic artery, actively controlled within the liver, and the portal vein, principally regulated by prehepatic resistance vessels. Portal flow may modulate the hepatic arterial circulation through the "hepatic arterial buffer response." Hypotension caused by hemorrhage causes a decrease in portal blood flow but hepatic arterial flow is maintained by autoregulation and by the "buffer" response. Hypothermia (28 degrees C) has little effect on hepatic arterial flow, but portal flow may increase. During cardiopulmonary bypass, total liver blood flow is better maintained at a pump flow rate of 2.4 than at 1.2 L/min/m2. Perfusion at 28 degrees C causes an increase in portal flow and a slight decrease in hepatic arterial flow. Total hepatic blood flow is better preserved during cardiopulmonary bypass at 1.2 L/min/m2 by pulsatile than by nonpulsatile flow; however, no significant difference was noted between pulsatile and nonpulsatile perfusion at a bypass flow rate of 2.4 L/min/m2, particularly at 28 degrees C. During cardiopulmonary bypass, hepatic blood flow is better maintained by high pump flow than by low pump flow rates. Hypothermic cardiopulmonary bypass may benefit the hepatic circulation, although the additional advantages usually gained by the use of pulsatile perfusion may be partly lost when hypothermia is combined with a high pump flow rate.