Between 1986 and 1994, hepatic local arterial infusion chemotherapy with implanted reservoir was performed for 51 patients with hepatocellular carcinoma (HCC) who were not indicated for surgery, PEIT and/or TAE because of the advanced stage of tumors and impaired liver function. We compared direct effects, survival rates, and rates of outpatients by dividing these 51 patients into 3 groups according to the methods of administration--repeated one-shot arterial infusion group, continuous arterial infusion group and intermittent high-dose arterial infusion group. There were no significant differences among these three groups in terms of background factors. The continuous infusion group had significantly better survival rates than the repeated one-shot infusion group, and the survival rate of the intermittent high-dose infusion group was similar to that of the continuous infusion group, although there were no significant differences in effectiveness among the three groups. Patients in the continuous infusion group and intermittent high-dose infusion group could receive almost all of their treatment as outpatients. Furthermore, more patients in the intermittent high-dose infusion group could receive whole treatment without hospitalization than patients in the continuous infusion group. Because intermittent high-dose arterial infusion of 5-FU showed about the same survival rate as continuous arterial infusion and because it can maintain high quality of life, it is suggested to be a standard method for local arterial infusion of 5-FU for severe advanced HCC.