To determine independent prognostic factors influencing the survival of patients with hepatocellular carcinoma (HCC) treated with transcatheter arterial chemoembolization (TACE).Ninety-one patients with unresectable HCC were treated with 269 repetitive TACE. The dosages of epirubicin (40-60 mg) and ethiodized oil (8-20 ml) were adjusted to tumor size and liver function. The impact of tumor size, macroscopic tumor type, tumor location, portal vein infiltration, capsular infiltration, tumor vascularization, uptake of ethiodized oil within the tumors, Child-Pugh-Class and Okuda-Stage on patient survival were evaluated by means of univariate and multivariate regression analysis.The following independent prognostic factors were found: tumor type (nodular vs. infiltrating, p = 0 008), tumor size (p = 0.01), Child-Pugh-Class (A vs. B; p = 0.02) and grade of tumor vascularization (p = 0.04). In 57 patients with HCC of the nodular type, the median survival time was significant longer than in 32 patients with HCC of the infiltrating type (17.0 months vs. 7.9 months; p < 0.003; 2 tumors could not be classified). The 1-, 2- and 3-year-survival rates were significantly higher in 57 patients with Okuda-Stage I disease, compared to 34 patients with Okuda-Stage II and III disease (73%, 31% and 8% vs. 23%, 6% and 4% p < 0.0001).Tumor type, tumor size and grade of liver cirrhosis have an independent impact on prognosis of patients with HCC treated by TACE. An appropriate selection of patients is necessary to improve patients survival.