Hepatocellular carcinomas that should not be treated with transarterial embolization (TAE) were treated with intermittent intraarterial infusion chemotherapy using a subcutaneously implanted reservoir. Among 37 patients, in whom objective evaluation of the initial response was possible, 3 cases of CR, 12 of PR, 10 of NC and 12 of PD were obtained. The response rate was 40.5%. Culminate 1 year survival rates were 80.0% for responders (CR + PR), 4.5% for non-responders (NC + PD) and 35.1% as an average for all patients. The survival in responder patients was better than in non-responders (p < 0.001). To determine the role of this therapy we divided the cases into two groups by anamnesis of TAE and investigated the initial response and prognosis based on the grade of portal vein invasion in each group. In one group of 18 patients, who had not been treated with TAE, there were no relationships between the grade of portal vein invasion and the initial response and between the grade of portal vein invasion and the prognosis. In this group the initial response was better in patients whose one shot infusion chemotherapy immediately before intermittent intraarterial infusion chemotherapy had been effective than in the others. In the other group of 19 patients, who had been treated with TAE, the initial response and prognosis were better in patients with lower grades of portal vein invasion (Vp0-2) than in those with the highest grade (Vp3).