AbstractThere is an increasing demand to integrate land surface temperature (LST) into climate research due to its global coverage, which requires a comprehensive knowledge of its distinctive characteristics compared to near‐surface air temperature (Tair). Using satellite observations and in situ station‐based data sets, we conducted a global‐scale assessment of the spatial and seasonal variations in the difference between daily maximum LST and daily maximum Tair (δT, LST − Tair) during 2003–2014. Spatially, LST is generally higher than Tair over arid and sparsely vegetated regions in the middle‐low latitudes, but LST is lower than Tair in tropical rainforests due to strong evaporative cooling, and in the high‐latitude regions due to snow‐induced radiative cooling. Seasonally, δT is negative in tropical regions throughout the year, while it displays a pronounced seasonality in both the midlatitudes and boreal regions. The seasonality in the midlatitudes is a result of the asynchronous responses of LST and Tair to the seasonal cycle of radiation and vegetation abundance, whereas in the boreal regions, seasonality is mainly caused by the change in snow cover. Our study identified substantial spatial heterogeneity and seasonality in δT, as well as its determinant environmental drivers, and thus provides a useful reference for monitoring near‐surface air temperature changes using remote sensing, particularly in remote regions.