In order to explore a dose distribution verification procedure of intensity modulated radia-tion therapy (IMRT) for nasopharyngeal carcinoma (NPC) and establish its evaluation criteria, we per-formed 35 two-dimensional (2D) patient-specific IMRT verifications over the year 2006. The percent of pixels passing ? and the normalized agreement test (NAT) index were mainly used to represent the agreement between the measured and computed dose distributions with three criteria (2%/2 ram, 3%/3 mm and 5%/3 ram) as recommended in the literature. The results were that all cases passed through verifications with three criteria except that the NAT index of one case was beyond the limitation, and the three tolerance levels of 2%/2 ram, 3%/3 mm and 5%/3 mm produced similar clinical verification re-sults but led to different percent of pixels passing γ and NAT index. Our data showed that the percent of pixels passing γ and the NAT index were complementary to evaluate future IMRT verifications as two significant metrics. Due to the influence of the noise and the trait of the software, we considered an IMRT plan as acceptable in case of the percent of pixels passing γ >95% and the NAT index <5 with the 5%/3 mm criteria for IMRT patient-specific quality assurance (QA).