In this study, in situ observations of turbulent heat flux and the OAFlux data are compared with three satellite-derived datasets, i.e., JOFURO (Japanese Ocean Flux Data Sets with use of Remote Sensing Observations), HOAPS-2 (Hamburg Ocean Atmosphere Parameters and Fluxes from Satellite data version 2), and GSSTF-2 (Goddard Satellite-Based Surface Turbulent Fluxes version 2), in the South China Sea (SCS). The three satellite-derived datasets have different features. In general, the spatial distributions and temporal variabilities of JOFURO and GSSTF-2 data are consistent with those of the OAFlux data. However, these two datasets largely underestimate the average latent and sensible heat fluxes over the entire basin. The former underestimates the heat fluxes by 10%-20% and the underestimation in the latter can reach up to more than 50%. The HOAPS-2 data are consistent with the in situ data, but there exist distinct differences in the temporal variability when compared with other datasets, especially the amplitude of seasonal variation and the phase of interannual variation in the sensible heat flux. The comparison reveals that the underestimation of the latent and sensible heat fluxes in the vicinity of Hainan Island and the southern SCS is the main reason for the underestimation in the entire SCS basin.