Arterial embolization of the liver may temporarily retard the growth of its primary and secondary tumors which are both mainly nourished arterially. Addition of radioisotopes, mostly (131)I or (90)Y, results in radioembolizations which predominantly act by radiation and less by ischemia. They may therefore be utilized in the absence of portal venous flow when conventional embolization is hazardous. (131)I-oily radioembolization seems to prolong short-term survival in such patients with unresectable hepatocellular cancers, and to improve the prognosis after resection of hepatocellular cancer. The procedure does however not palliate better than "cold" chemoembolization in patients with preserved portal flow, except for having milder side effects. Embolization with (90)Y-coupled microspheres may shrink primary and secondary liver tumors but has so far unproven effects on survival. Embolization of portal venous branches gives compensatory hypertrophy of the non-embolized liver and can increase the volume of the future remnant liver before resection. This diminishes the risk for postoperative liver failure after extensive resection and/or in the presence of chronic liver disease, and permits wider surgical indications. Tumor growth may however be accelerated, and the hypertrophy is inhibited by severe liver parenchymal disease in which situation the method would be most needed. Experimental use of liver arterial embolizations includes combined arterial and portal embolizations, i.e. "chemical hepatectomy," arterial embolizations before external radiotherapy, administration of boron for neutron capture therapy, immunoembolizations, and future gene therapy.