Long-term snow-depth observations at Arxan station are used to evaluate the application of snow-depth products retrieved from the AMSR-E and AMSR-2 microwave data and Chinese snow-depth dataset developed by Chinese researchers and to establish a new snow-depth retrieval algorithm. The statistical snow-cover days and maximum snow-depth records derived from the 35-year Chinese snow-depth dataset and station observations are consistent, particularly after 2000. The snow-depth variation trend estimated from the snow-depth products retrieved from the AMSR-E and AMSR-2 microwave data is consistent with that retrieved from the station observations, with the correlation coefficient greater than 0.6. However, the variation range of the snow-depth products is wider than that of the station observations. Thus, the root mean square error (RMSE) of both snow-depth datasets is high (i.e., approximately 13 cm). The Chinese snow-depth dataset at Arxan station shows a higher correlation coefficient of 0.65 and a lower RMSE of 6.3 cm than the station observations. To better estimate snow depth in the Arxan region, a new snow-depth retrieval method is developed using both space-borne passive microwave brightness temperature and observed snow-depth data at Arxan station. The validation shows that the snow-depth data retrieved using the new method has a higher correlation with the observations (i.e., approximately 0.77) and a lower RMSE (i.e., approximately 4.68 cm) than the snow-depth products retrieved from the AMSR-E and AMSR-2 microwave data and Chinese snow-depth dataset used in this study.