To enhance the understanding of airflow characteristics in the human respiratory system, the expiratory airflow in a human respiratory tract model was simulated using large eddy simulation and dynamic mesh under different expiration conditions aligned with clinically measured data. The airflow unsteadiness was quantitatively assessed using power spectral density (PSD) and spectral entropy (SE). The following findings were obtained:​ (1) The airflow is highly turbulent in the mouth-pharynx region during expiration, with its dynamic characteristics being influenced by both the transient expiration flow pattern at mouth piece and the glottis motion. (2) PSD analysis reveals that the expiratory airflow is very unsteady, exhibiting a broad-band attenuation spectrum in the pharynx-trachea region. When only transient expiration or glottis motion is considered, the PSD spectrum changes slightly. When both are ignored, however, the change is significant, with the peak frequency reduced to 10% of the real expiration condition. (3) SE analysis indicates that the airflow transitions into turbulence in the trachea, and there may be multiple transitions in the region of soft palate. The transient expiration or glottis motion alone increases turbulence intensity by 2%–15%, while ignoring both reduces turbulence intensity by 10%–20%. This study implies that turbulence characteristics can be significantly different under different expiratory conditions, and therefore it is necessary to determine the expiratory flow characteristics using clinically measured expiratory data.