Recently, the China Meteorological Administration (CMA) released a new Global Atmospheric Reanalysis (CRA-40) dataset for the period 1979-2018. In this study, surface relative humidity (RH) from CRA-40 and other current reanalyses (e.g., CFSR, ERA5, ERA-Interim, JRA-55, and MERRA-2) is comprehensively evaluated against homogenized observations over China. The results suggest that most reanalyses overestimate the observations by 15%-30% (absolute difference) over the Tibetan Plateau but underestimate the observations by 5%-10% over most of northern China. The CRA-40 performs relatively well in describing the long-term change and variance seen in the observed surface RH over China. Most of the reanalyses reproduce the observed surface RH climatology and interannual variations well, while few reanalyses can capture the observed long-term RH trends over China. Among these reanalyses, the CFSR does poorly in describing the interannual changes in the observed RH, especially in Southwest China. An empirical orthogonal function (EOF) analysis also suggests that the CRA-40 performs better than other reanalyses to capture the first two leading EOF modes revealed by the observations. The results of this study are expected to improve understanding of the strengths and weaknesses of the current reanalysis products and thus facilitate their application.