In this paper we analyze the results of dose map verifications for patient's IMRT (Intensity-modulated radiation therapy) plans and study the factors that may influence the accuracy of verification. MapCHECK, a two-dimensional diode array, was used to measure the dose maps for 1242 plans (14540 fields) from May 2004 to August 2008. The plans were designed with Pinnacle³ planning system. The passing rate of beams was determined when the acceptance criterion was 2%/2 mm, 3%/3 mm and 4%/4 mm. And the data with 3%/3 mm criteria was analyzed in more detail. The considered factors were beam modeling, optimization mode and treatment site. The median passing rate of total fields was 93.5%, 98.8%, and 100% when the acceptance criterion was 2%/2 mm, 3%/3 mm and 4%/4 mm, and the interquartile range were 11.1%, 3.8%, and 1.3%, respectively. The results of direct machine parameter optimization (DMPO) planning mode was better than those of multiple-step mode and beam modeling of new accelerators was better than that of old accelerators. These indicate that beam modeling is the key point of improving passing rate of IMRT verification and the influence of treatment site was little. The factors, the total number of segments, minimum area of segments and minimum monitor unit (MU) of segments, also influence the dosimetric accuracy of IMRT plan verification.