OBJECTIVE. The purpose of our study was to assess the efficacy of transcatheter arterial chemoembolization in the treatment of hepatocellular carcinoma using two different infusion schedules. MATERIALS AND METHODS. Chemoembolization with lipiodol-mediated injection of doxorubicin was performed in 160 patients. In the first group of 80 patients, conventional chemoembolization was initially planned to be repeated at least three times at 2-month intervals. In the second group of 80 patients, chemoembolization was used selectively and repeated only when necessary on the basis of follow-up CT or MR imaging. According to the Okuda classification, 72 patients were stage 1, 33 belonging to group 1 (subgroup 1, Okuda I) and 39 belonging to group 2 (subgroup 2, Okuda 1). Eighty-eight patients were stage 2, 47 belonging to group 1 (subgroup 1, Okuda 2) and 41 belonging to group 2 (subgroup 2, Okuda 2). RESULTS. Complications of transcatheter arterial chemoembolization occurred in 19 patients from group 1 and six patients from group 2 (p < .001). The mean time between the first and the third courses was significantly different between group 1 (4 months) and group 2 (14 months) (p < .001). The 1-year, 2-year, and 3-year survival rates were significantly different between subgroup 1, Okuda 1, (58%, 28%, 11%) and subgroup 2, Okuda 1 (89%, 68%, 39%) (p < . 001), and between subgroup 1, Okuda 2 (19%, 0%, 0%), and subgroup 2, Okuda 2 (48%, 31%, 15%) (p < .001). CONCLUSION. The efficacy and tolerability of chemoembolization increase when it is used selectively and repeated only when necessary. Such technical considerations might explain some of the discrepancies of the results of chemoembolization in published data.