From January 1986 to December 1988, a prospective trial of transcatheter arterial treatment was carried out for hepatocellular carcinoma (HCC). Two hundred seventy-five patients were included. Okuda's staging system was employed. Patients with Stage I and II HCC were treated by transcatheter arterial embolization (TAE) with a gelatin sponge containing an anti-cancer agent (protocol 1a); a gelatin sponge and iodized oil mixed with an anti-cancer agent (protocol 1b); or iodized oil mixed with an anti-cancer agent (protocol 2). Patients with Stage III HCC were treated with iodized oil with anti-cancer agent (protocol 2). As an exception, patients with an unsuccessful superselective catheterization into the proper hepatic artery by Seldinger technique or obstruction of the main trunk of the portal vein were treated with percutaneous transcatheter arterial infusion into the common hepatic artery regardless of stage (protocol 3). Tumor type and extension, area of tumor involvement, portal vein involvement, method of treatment, and presence of ascites and icterus were found to be the significant factors for an initial response to therapy. Treatment method was the most important factor. Respective survival rates at 1 and 2 years were 70.9% and 55.3% for protocol 1a; 62.3% and 43.8% for protocol 1b; 37.8% and 18.3% for protocol 2; and 16.5% and 0% for protocol 3. Many factors proved to significantly influenced prognosis; however, tumor type had the most important prognostic significance followed by AFP value, ascites, treatment protocol, and area of tumor involvement.