Abstract: Aiming at the problem that the common mode error analysis of the vertical coordinate time series of the Global Navigation Satellite System (GNSS) in Northwest China is not comprehensive and in-depth, the MSSA algorithm is used as the theoretical basis to preprocess the data of 61 stations in the three northwestern provinces (Xinjiang, Gansu, Qinghai) to improve the accuracy and integrity of the data, and on this basis, the common mode error of GNSS coordinate time series in the region is extracted and analyzed. The characteristic source of the Common Mode Error (CME) was investigated by correlation analysis between the extracted CME and the displacement time series caused by Hydrologic Load (HYDL), Non-Tidal Atmospheric Load (NTAL) and Non-Tidal Ocean Load (NTOL). Compared with the sequence before and after the removal of the common mode error, the average accuracy of the residual standard deviation is increased by 27.5%, and the maximum reduction of the station is reduced by 66.7%. The results show that the common mode error in this region is significantly affected by three kinds of surface mass loads: hydrologic load, non-tidal atmospheric load and non-tidal ocean load. Through the extraction and analysis of common mode errors and the research on the source of error characteristics, the GNSS coordinate time series accuracy of the three northwestern provinces can be further improved, so as to provide higher precision data support for seismic displacement and crustal deformation research.