Background: The purpose of this study is to compare SOURCE 8 and 20 in the EGSnrc-based DOSXYZnrc Monte Carlo code for Jaws-only intensity-modulated radiation therapy (JO-IMRT) dose distribution, and demonstrate the advantage of SOURCE 20 to SOURCE 8 to treat head-and-neck cancer. Materials and Methods: The clinical photon beams of the HPD Siemens Primus linear accelerator simulated using the BEAMnrc code and then verified by measurement. The phase-space files generated by the BEAMnrc code were used as an input for the DOSXYZnrc to calculate the JO-IMRT dose distributions of patients (in form of CT images) using the SOURCES 8 and 20. The isodose distribution on slices, DVH and gamma index (3%/3 mm, 2%/2 mm, and 1%/1 mm) were compared with Monte Carlo and treatment planning system (TPS) results. Furthermore, the efficient computation of dose distributions such as time running, working load and uncertainty error calculation also considered for evaluation. Results: JO-IMRT dose distributions calculated by SOURCE 8 and 20 compared with the Prowess Panther TPS. The time running, working load based on SOURCE 20 were significantly less than SOURCE 8. Futhermore, the uncertainty calculation based on SOURCE 20 was also significantly less than SOURCE 8.Conclusion: The JO-IMRT dose distributions calculated by SOURCE 20 are more efficient than SOURCE 8.