Objective To compare the concordance among RECIST, EASL and modified RECIST criteria for the evaluation of tumor response after transarterial chemoembolization of primary liver cancer.Methods Fifty patients with primary liver cancer underwent 2 TACE cycles separated by 30-40 days.Triphasic helical CT or MRI scans were performed at baseline, at 4 weeks after TACE procedure, and 2 independent radiologists evaluated tumor response according to above-mentioned three different criteria. Chisquare test was used to compare the response rate, and kappa coefficients were used to evaluate the coherence. Results When tumor responses were evaluated using the RECIST-EASL and modified RECIST criteria, the numbers of the patients achieved complete response, partial response, stable disease,progressive disease were 0, 10, 30, 10; 6,21,14,9; 6,21,13,10 respectively. The objective response rates for three different criteria were 20%, 54%, 54% respectively ( P ＜ 0. 01 ). Kappa coefficients between RECIST and EASL, between RECIST and modified RECIST, between EASL and modified RECIST were 0. 382, 0. 170, and 0. 857 (P = 0. 000). Conclusions RECIST criteria underestimates the extent of tumor response after TACE in primary liver cancer. Both EASL and modified RECIST criteria appear to agree with each other in determining treatment response. Furthermore, the modified RECIST is more convenient in clinical practice compared with EASL criteria.