This paper addresses the directional land surface emissivity (LSE) retrieval from the data acquired by the MEdium Resolution Spectral Imager (MERSI) on Fengyun 3D and 3E (FY-3D/E) and the Moderate-resolution Imaging Spectroradiometer (MODIS) on Terra and Aqua. First, a method, including LSE retrieval and multi-satellite data assimilation, is developed based on radiative transfer theory. Then, the MERSI/FY-3D/E and MODIS/Aqua data are transferred to the effective MODIS/Terra measurements under the same observation conditions. Next, the bi-directional reflectances in MODIS/Terra channel 20 are derived based on the temperature independent spectral indices (TISI) concept. Later, the directional LSEs in the channel 20 are retrieved using the RossThick-LiSparse-R model and the Kirchhoffâ\euro™s law. After that, the directional LSEs in MODIS/Terra channels 29, 31 and 32 are retrieved. Finally, the retrievals are validated with the MODIS/Terra land surface temperature and emissivity (LST/E) 8-day level 3 (MOD11C2) product and the in situ data at three bare area sites. The results show that the multi-satellite data provide more information of view angles and solar angles, which makes the retrievals more robust. Over the vegetated areas with normalized difference vegetation index (NDVI) in (0.3, 1.0), the retrievals in this work are basically consistent with the MOD11C1 product, and the LSE differences are -0.037±0.032, -0.013±0.019, -0.020±0.014 and -0.028±0.020 in the channels 20, 29, 31 and 32, respectively. Over the bare areas with NDVI in (0.0, 0.3], large differences exist between the retrievals in this work and the MOD11C2 product, especially for the channel 20. However, the retrievals over the bare areas in this work agree with the in situ LSEs with mean absolute errors of 0.026, 0.060, 0.007 and 0.012 in the four channels, respectively. In general, the retrievals from the multi-satellite data in this work are valid and accurate.