Although portal dosimetry is used to provide quality assurance (QA) for intensity-modulated radiation therapy (IMRT) treatment plans, trends in agreement between the portal dose prediction (PDP) and the measured dose have not been clarified. In this work, we evaluated three scalar parameters of agreement for 152 treatment plans (1152 treatment fields): maximum gamma (gamma max), average gamma (gamma avg), and percentage of the field area with a gamma value greater than 1.0 (gamma % > 1). These data were then used to set clinical action levels based on the institutional mean and standard deviations. We found that agreement between measured dose and PDP was improved by recalculating the fields at lower dose rates. We conclude that action levels are a useful tool for standardizing the evaluation of EPID-based IMRT QA.