The clearance from the surface of the liver of krypton 85 following its injection into the portal vein was measured in 10 patients. Six control patients (i.e. those not suffering from portal hypertension) averaged flow values of 142 ml per 100 g liver weight per min. A repeat flow estimation in these patients was not significantly different. Four patients with portal hypertension were studied. In 2 with congenital hepatic fibrosis the portal vein was temporarily occluded to allow prediction of the degree of hepatic perfusion likely to be achieved by the hepatic artery alone after portal deviation. The flow after portal venous clamping was unaffected in one of these cases but reduced by 45 per cent in the other. In a patient undergoing a Warren distal splenorenal shunt there was no change in flow after the procedure. If measurements are performed before and during temporary portal vein occlusion, an assessment of the flow reduction to be expected after a shunt procedure can be made. We suggest that a knowledge of liver tissue perfusion might influence the choice of operative procedure.