Obstructive sleep apnea (OSA) is an airway disease caused by periodic collapse of the airway during sleep. Imaging-based subject-specific computational fluid dynamics (CFD) simulations allow non-invasive assessment of clinically relevant metrics such as total pressure loss (TPL) in patients with OSA. However, most of such studies use static airway geometries, which neglect physiological airway motion. This study aims to quantify how much the airway moves during the respiratory cycle, and to determine how much this motion affects CFD pressure loss predictions. Motion of the airway wall was quantified using cine MRI data captured over a single respiratory cycle in three subjects with OSA. Synchronously-measured respiratory airflow was used as the flow boundary condition for all simulations. Simulations were performed for full respiratory cycles with 5 different wall boundary conditions: (1) a moving airway wall, and static airway walls at (2) peak inhalation, (3) end inhalation, (4) peak exhalation, and (5) end exhalation. Geometric analysis exposed significant local airway cross-sectional area (CSA) variability, with local CSA varying as much as 300%. The comparative CFD simulations revealed the discrepancies between dynamic and static wall simulations are subject-specific, with TPL differing by up to 400% between static and dynamic simulations. There is no consistent pattern to which static wall CFD simulations overestimate or underestimate the airway TPL. This variability underscores the complexity of accurately modeling airway physiology and the importance of considering dynamic anatomical factors to predict realistic respiratory airflow dynamics in patients with OSA.