A total of 267 patients with hepatocellular carcinoma underwent ultrasound-guided radiofrequency ablation (RFA) in Peking University School of Oncology between 1999 and 2005 (421 RFA sessions). Among them, 254 patients were candidates for RFA treatment and the selective criteria were: (1) the greatest diameter of tumor <or=8.0 cm; (2) tumor number <or=4; (3) no obvious invasion into adjacent structures; (4) absence of extensive tumor thrombus in large vessels; (5) prothrombin time ratio greater than 60% and platelet count greater than 50x10(9)/L. Of the 256 HCC patients, 206 were male and 50 were female (mean age, 56.8 years; range, 24-80 years). The mean size of the tumors was 3.8 cm (range, 1.2-8.0 cm). Among the 256 HCC patients, 207 (80.8%) were not amenable for surgery due to impairment of liver function, post-operative recurrence, multiple tumors, senile, serious cardiac/respiratory damage or diabetes. According to the UICC-TNM staging system, 61, 90, 45, and 8 patients were in stages I, II, III, and IV, respectively. Fifty-two patients had recurrent HCC after surgical resection. Of these 256 HCC patients, their Child-Pugh grades of A, B and C were 150, 94, and 12, respectively. Of all the subjects, 151 patients had solitary tumors and 105 had multiple tumorsì and 65, 127, and 64 patients had tumors sized <or=3 cm, 3.1-5 cm, and >5 cm, respectively. According to tumor size, shape and location, we adopted a defined treatment strategy, which consisted of a mathematical protocol, an individualized protocol and adjunctive measures. And several methods were also used to prevent and deal with complications in tumors with different features. In this series the tumor complete necrosis rate (CR)was 95.2% (356/374 tumors). It was higher in <or=3.5 cm tumors with a CR of 98.5% (200/203 tumors) than in > 3.5 cm tumors with a CR of 91.3% (156/171 tumors). CR were 95.6% (44/46 tumors) for tumors near the gallbladder, 92.9%(79/85 tumors) for tumors near the diaphragm, 90.9%(40/44 tumors) for tumors near the gastrointestinal tract, 91.2% (31/34 tumors) for tumors near large vessel. In a follow-up period of 2-69 months, the local recurrence rates were 11.7% for HCC and 12.5% for recurrent HCC. The incidence of complications was 2.4% (10/409 sessions), including intraperitoneal hemorrhage (n=2), biliary duct stricture (n=1), hemothorax (n=1), bowel perforation (n=1) and needle tract seeding (n=5). Of these cases, only 3 required operation and the mortality related to RFA was zero in this series. We used Kaplain-Meier method and log-rank test to estimate and compare the survival rate. The 1-, 3-, and 5-year survival rates after RFA were 83.3%, 66.9%, 41.2%, respectively for all HCC patients and 74.6%, 41.3%, 33.6%, respectively for recurrent HCC. Survivals based on TNM stage, Child-Pugh grade, tumor number and tumor size are shown in Table 1. In conclusion, RFA with standard protocol has evolved as a minimally invasive local treatment that could achieve satisfactory outcomes for small liver tumors, and has become an effective and relatively safe alternative for the treatment of advanced tumors and recurrent tumors, which are not suitable for traditional therapy. RFA has broaded the treatment threshold for hepatic malignancies and might become one of the regular treatment methods in focal liver tumor and find wide application.