To determine the route of prostaglandin E1 (PGE1) administration which would have the greatest protective effect against hepatic warm ischemia, two experiments were performed using dogs. The pharmacokinetics of PGE1 were investigated in a preliminary study, after which, the effects of PGE1 in a 90-min warm ischemic liver model were examined. The dogs were divided into three groups of ten, according to the treatment given: group A was an untreated control group, group B received PGE1 intravenously, and group C received PGE1 intraportally. The PGE1 was infused continuously at a rate of 0.02 microliters/kg/min before and after ischemia. All the dogs in groups A and B died within 24 h of induced ischemia. Whereas, six of the ten dogs in group C survived for over 3 days. The arterial ketone body ratio was not maintained in groups A and B, but it was in group C. Furthermore, in group C the serum lipid peroxide level, which reflects hepatocellular membrane damage, was maintained at a lower level than that in the other groups after ischemia. Electron microscopy revealed sinusoid destruction and changes in both the plasma membrane and parenchymal cell mitochondria in groups A and B, while in group C these structures were well preserved. These findings confirmed that intraportally administered PGE1 improved the hepatic microcirculation and stabilized the hepatocellular membranes. Our results indicate that intraportal administration of PGE1 has a greater protective effect than intravenous administration against warm ischemic liver injury.