OBJECTIVE: To evaluate dose differences predicted between using Anisotropic Analytical Algorithm (AAA) and Acuros XB (AXB) in patients diagnosed with locally advanced non-small cell lung cancer (NSCLC) treated with intensity modulated radiation therapy (IMRT). METHODS: A phantom study was done to evaluate the dose prediction accuracy of AXB and AAA beyond low-density medium by comparing the calculated measurement results. Thirty-two advanced NSCLC patients were subjected to IMRT. The dose regimen was 60 Gy over 30 fractions. Effects on planning target volume (PTV) and organ-at-risk (OAR) were evaluated. Clinically acceptable treatment plans with AAA were re-calculated using AXB algorithms with two modes Dw and Dm at the same beam arrangements and multileaf collimator leaf settings as with AAA. RESULTS: Using AXB yielded better agreement with the measurements and the average dose difference for all points was about 0.5%. Conversely, using AAA showed a larger disagreement with measured values and the average difference was up to 5.9%. The maximum relative difference was between AXB_Dm and AAA for PTV dose (D98 %). The percentage dose differences of plans calculated by AAA, AXB_Dw and AAA, AXB_Dm revealed that AAA overestimated the dose than AXB. Regarding OAR, results showed significant difference for lungs-PTV. CONCLUSIONS: AXB algorithm yields more accurate dose prediction than AAA in heterogeneous medium. Differences in dose distribution are observed when plans re-calculated with AXB indicating that AAA apparently overestimates dose, particularly the PTV dose. Thus, AXB algorithm should be used in preference to AAA for cases in which PTVs are involved with tissues of highly different densities, such as lung.