Cloud radiative effects (CREs) at the top of the atmosphere (TOA) in three reanalysis datasets (the European Center for Medium-Range Weather Forecasts Interim Reanalysis (ERA-Interim), the Japanese 55-yr Reanalysis Project (JRA-55), and the Modern-Era Retrospective Analysis for Research and Applications (MERRA)) are evaluated using recent satellite-based observations. The reanalyses can basically capture the spatial pattern of the annual mean shortwave CRE, but the global mean longwave CRE in ERA-Interim and JRA55 is weaker than observed, leading to overestimations of the net CRE. Moreover, distinct CRE biases of the reanalyses occur in the Intertropical Convergence Zone (ITCZ), coastal Pacific and Atlantic regions, and East Asia. Quantitative examination further indicates that the spatial correlations of CREs and TOA upward radiation fluxes with corresponding observations in ERA-Interim are better than in the other two reanalyses. Although MERRA has certain abilities in producing the magnitudes of global mean CREs, its performance in terms of spatial correlations in winter and summer are worse than for the other two reanalyses. The ability of JRA55 in reflecting CREs lies between the other two datasets. Compared to the global mean results, the spatial correlations of shortwave CRE in East Asia decrease and the biases of regional mean CREs increase in the three reanalyses. This implies that, currently, it is still difficult to reproduce East Asian CREs based on these reanalyses. Relatively, ERA-Interim describes the seasonal variation of East Asian CREs well, albeit weaker than observed. The present study also suggests that in-depth exploration of the ability of reanalysis data to describe aspects relating to cloud properties and radiation is needed using more comprehensive observations.