Herein, the prediction accuracy of three tropospheric refractive index profile models commonly used at home and abroad was analyzed using the measured data of the tropospheric zenith delay and meteorological environment obtained from 11 international global positioning system service stations in China. Moreover, the applicability of the models was evaluated in high altitude areas and with seasonal changes. Results show that the exponential and segmented models achieve a higher accuracy at high altitude areas, while the Hopfield model shows greater advantages in areas other than high altitudes. Generally, the accuracy of the segmented and Hopfield models is higher in spring and winter and lower in summer and autumn. The accuracy of the exponential model shows a negligible relation with the season. The findings of this study have important reference values for the application research of the tropospheric refraction error correction model.