This paper proposes a comprehensive quality control (CQC) algorithm for the Chinese ground-based navigation satellite system (GNSS) water vapor products. The CQC algorithm consists of two sections: quality checks and comprehensive decision-making algorithm. The quality checks consist of 7 parts: limit check to eliminate errors that exceed reasonable limits, buddy check and low-pass filter check for better time consistency, neighboring station check, anomaly check and peak-valley value check for better spatial consistency, and background check to mark out data deviate from background field for assimilation application. After each check, the data that exceeds the threshold is marked, and then the comprehensive decision-making algorithm is used to score the marked data, and finally flag (correct, suspicious, or error) the data. Based on the quality-controlled observation data, the precipitable water vapor simulation of five sets of reanalysis data, including China"s first-generation global atmosphere reanalysis (CRA) product, were evaluated. The results show that the simulated total water vapor of all the reanalysis data in winter is slightly higher than the observation in winter and significantly lower than the observation in summer. Spatially, the simulated water vapor content in southern and western China is lower than the observation and this situation is more obvious in the summer half of the year. Relative to the observation, the average bias (B-O) of CRA is -0.633mm, and the root mean square error is 3.650mm. The deviation of CRA relative to observation is slightly lower than ERA-Interim but slightly higher than ERA5, which is significantly better than JRA55 and NCEP2 results.