Abstract: In order to successfully apply the Becker and Li’s split window approach, which was proposed based on the NOAA-9 AVHRR data, to estimate the Land Surface Temperature （LST） from the different AVHRRs and VIRR data and further to provide a high-precision, long-time, and high resolution LST dataset for climate change research, a modified Becker and Li’s split widow approach is developed based on the radiative transfer equation in this paper. To begin with, the MODTRAN 4.1 is used to generate the spectral radiance data under a variety of surface and atmosphere conditions. Then, the temperature dataset (TS,T4,T5) is built by convolving the spectral radiance data with the spectral response functions of channels 4 and 5 of the AVHRRs and VIRR. The parameters of the Becker and Li’s split window approach are subsequently recalculated based on the temperature dataset using the least squares regression method. As an example of validation, a single image of the AVHRR-17 over the Beijing area acquired at 03:12 UTC 27 April 2008 was used to retrieve the LST using the modified Becker and Li’s approach, and the comparison between the retrieved LST from AVHRR-17 and that from the MODIS, which was provided by the University of Tokyo in Japan, over the same region at the same time shows that the correlation coefficient is 0.88 and the root mean square deviation (RMSD) is 2.1 K. Furthermore, the frequency histogram of the LST difference image shows that about 69.6% and 37% pixels in the image have the values falling within the scope of ±2 K and ±1 K, respectively.