Abstract: Based on the hourly surface (2 m) temperature and humidity data of 2364 national meteorological stations during 2021−2023, the applicability of ERA5, CRA, and ART reanalysis temperature and humidity datasets in China is preliminarily compared and evaluated. The results show that the three reanalysis datasets can reproduce the spatial and temporal distribution characteristics of surface air temperature and humidity, but there are some differences in the reproduction results. The CRA datasets is warmer than the observed values, whereas the ERA5 datasets is colder. The humidity in all three datasets is lower than the observed humidity. Among them, the ART datasets has the highest reliability, followed by the CRA datasets. In terms of spatial distribution, the credibility of the three reanalysis datasets is relatively high in South China, East China, North China, Southwest China, and Northeast China. However, it is relatively low in the Northern semi-arid regions, Northwest China, Xinjiang, and the Qinghai-Xizang Plateau. The ART datasets performs best in the Qinghai-Xizang Plateau, South China, East China, Southwest China, North China, Northeast China, Northern semi-arid regions, and Northwest China, followed by the CRA datasets. In the Xinjiang region, the ART datasets performs best, followed by the ERA5 datasets. In the time dimension, the three datasets are better in summer than in winter.