Feng-Yun (FY-3) is the second generation of the Chinese Polar Orbiting Meteorological Satellites with global, three-dimensional, quantitative, and multispectral capabilities. Medium Resolution Spectral Imager (MERSI) has 20 channels onboard the FY-3A and FY-3B satellites, including five channels (four VIS and one thermal IR) with a spatial resolution of 250m. The top of the atmosphere signal are necessary to be radiometrically calibrated and corrected for atmospheric effects based on surface reflectance, especially in land surface remote sensing and applications. This paper presents an atmospheric correction algorithm for FY3/MERSI data over land in China, taking into account the directional properties of the observed surface by a kernel-based Bi-directional Reflectance Distribution Function (BRDF) model. The comparison with MODGA and ASD reflectance showed that there is a good agreement. Therefore, FY3/MERSI can serve a reliable and new data source for quantifying global environment change.