Experiments were carried out to determine the accuracy and validity of estimations of hepatic blood flow from clearance data during infusions of galactose in anesthetized cats. Clearance calculations were compared directly with the measured hepatic blood flows using a hepatic venous long-circuit technique. This technique allowed direct measurement and alteration of hepatic blood flow and collection of arterial and mixed hepatic venous blood samples without depletion of the animal's blood volume. It was found that infusions of galactose could not be used to estimate accurately hepatic blood flow. Infusion rate could not be used as an estimate of hepatic or splanchnic uptake owing to substantial and variable extrasplanchnic uptake. As a result, estimated hepatic flows allowing for incomplete extraction overestimated the true flow. On the other hand, extraction was less than 100%. This caused systemic galactose clearance to underestimate hepatic blood flow. These errors could cancel each other giving an apparently good estimate of hepatic flow from systemic galactose clearance. This agreement was fortuitous and occurred only at a specific dose and blood flow. We conclude that in the absence of independent measurements of both extrasplanchnic uptake and splanchnic extraction of galactose, systemic galactose clearance is not a reliable measure of hepatic blood flow in anesthetized cats. Until proved otherwise, it seems likely that this is also true in humans.