Perfusion of a dog liver was carried out with a circulating liver-perfusion apparatus, provided with two blood supplies from the artery and the portal vein, and the effect of various drugs on the changes of arterial and portal inflow, hepatic venous outflow, and the liver volume were recorded. From these recordings, sites and modes of action of drugs on blood vessels of the liver were examined. It was observed that there was a partially reciprocal relationship between the inflow of hepatic artery and of the portal vein. Adrenaline chiefly caused dilating action on the hepatic veins in a small dose, but when the dose was comparatively large, constriction of the arterial and portal branches being more predominant. However, the hepatic veins were opened in some liver preparations even when the dose of adrenaline was fairly large. The action of noradrenaline seemed to be fundamentally similar to that of adrenaline. The action of ephedrine and tyramine was similar to that of adrenaline though much weaker and more lasting. Histamine itself caused constriction of the hepatic veins but its action on blood vessels in other parts was much weaker. A small dose of adrenaline completely counteracted the histamine action on the hepatic vein. The actions of peptone and sinomenine were similar to that of histamine and it was assumed that the actions appeared through histamine release from the liver tissues. Under the experimental conditions employed, the action of sinomenine seemed to be stronger. The principal action of acetylcholine was constriction of the hepatic veins, which was far stronger than that reported by the previous workers on this animal, although this action was transitory and a secondary dilatation of the hepatic veins often followed. The action of pilocarpine was of the same type as that of acetylcholine but far weaker. A small dose of atropine suppressed these actions and itself dilated the hepatic veins. These observations strongly suggest the presence of the parasympathetic control of the liver circulation which had been overlooked to date. Barium salts effected strong constriction of the tributaries of the hepatic artery and this action was removed by papaverine. Strophanthine showed a histamine-type action and the action of pituitrin was also similar, though the action of pituitrin was sometimes similar to that caused by a small amount of adrenaline. Of the drugs acting chiefly on the hepatic veins, the comparatively stable histamine showed about the same effect by the arterial and portal injections, but the effect of adrenaline and acetylcholine which are unstable was far weaker when given into the portal vein. These facts suggest the presence of direct communications from the hepatic artery to the vein, i.e. the transhepatic arterial branches.