Background: This study was conducted to assess the accuracy of dose calculation near the air-phantom interface of a heterogeneous phantom for Acuros XB (AXB) and Anisotropic Analytical Algorithm (AAA) algorithm of a           6-MV flattening-filter-free beam, compared with film measurements. Materials and Methods: A phantom included air gap was specially manufactured for this study. In order to evaluate the dose near air gap-phantom interface, Eclipse treatment planning system equipped both AXB and AAA was used for the dose calculations. Measurements in this region were performed with radiochromic film. The central-axis dose (CAD) and off-axis dose (OAD) between calculations and measurements were analyzed for various field sizes and air gaps. The root-mean-square-error (RMSE) was used to evaluate the difference between the calculated and measured OAD. In order to quantify agreement between the calculated and measured dose distributions, the gamma analysis was performed with the 2%/2 mm and 3%/3 mm criteria. Results: For all fields traveling through 1 and 3 cm air gap, the maximum difference in the calculated CAD was -5.3% for AXB and 214.8% for AAA, compared to the measured CAD. For the RMSE between the calculated and measured OAD, the calculated OAD using AXB showed interval in the RMSE (from 4.4 to 12.7) while using AAA indicated broad (from 7.7 to 101.0). In addition, the gamma passing rates showed that AXB was higher agreement than AAA. Conclusion: This study demonstrated that AXB was more accurate in heterogeneous media near air-phantom interface than AAA when comparing the measured data.