Abstract: Snow piling and melting dynamically couple with the hydrological and climatological. It's very important to obtain the accurate snow parameters in time for the study of hydrology and climatology. It is also useful in mitigating or preventing from snow disaster. Microwave remote sensing has the advantage to penetrate the cloud, which is superior to the optical sensor data. With the microwave data from the satellite, it could better identify the snow, even those covered by cloud, or with the higher solar zenith angle. Not only snow cover, but also snow depth and snow water equivalent (SWE) could be obtained. SSM/I is a 7-channel 4-frequency radiometer whose frequencies are 19.35, 22.235, 37.0 and 85.5 GHz respectively. All frequencies are received in dual polarization (V and H) except 22 GHz (V). So, it is very suitable for snow parameters' retrieval. Since 1987, SSM/I data nearly 20 years are collected. Meanwhile, several global snow identifying methods are developed with these data. But it is found that the snow cover is overestimated, when compared with the operational visible-infrared products of the China National Satellite Meteorological Center (NSMC), especially over the Qinghai-Xizang Plateau. Similar conclusion is drawn by Basist. Therefore, it is necessary to develop an advanced method to improve the accuracy of snow cover identification. Using SSM/I data, an improved method is proposed to identify the snow cover in China and its adjacent areas (17°—57°N, 65°—145°E).Snow, rain, cold desert and frozen ground are the scattering matters, and have the similar signatures in the microwave band. That means either the maximum one of (T_B22V-T_B85V) and (T_B19V-T_B37V), or both of them are not less than 5 K. With the snow measurements from the ground observation stations in China and the operational snow cover products of the NSMC, the cause of overestimating snow cover of the old global methods is analyzed, and the conclusion is drawn that parts of the scattering matters, especially the frozen ground, are incorrectly identified as snow, which is the main cause of snow cover overestimated in Qinghai-Xizang Plateau and Mongolia with SSM/I data. Based on the foreign global snow identifying methods and the statistic analysis around six years' measurements from the observation stations in Inner Mongolia about snow, rain and frozen soil matching with SSM/I data, an advanced method depending on five identifying parameters is put forward: a new parameter, (T_B22V-T_B85V)-(T_B19V-T_B37V), is added, which can effectively reduce the influence of the frozen ground and the other scattering matters in snow identification. Finally, the results obtained using the improved method are validated with snow measurements from ground observation stations and operational snow cover products of NSMC in two aspects: one is the temporal variation, the other is the spatial distribution of snow cover. It is found that the improved method is more accurate in snow identifying than the existing global methods.