In order to obtain the important land surface ecological parameter—land sur face temperature (LST) quickl y, macroscopically, and to avoid the more parameters estimation and the limitati on of the parameters applicability range of the splitwindow algorithm, a stati stical model to retrieve LST was built up by using 13 clear air earth surface MO DIS data and 24 autoweather station data in 2005-2007. The new model can enhan ce the calculation speed and make more efficient use of the “threestations an d four nets” products built by China Meteorological Administration. The results indicate that after introducing the related parameters, the model relevance and accuracy were improved obviously, the combination of the water vapor channel of MODIS with the enhanced vegetation index (EVI) can obtain an optimal effect. Compared with the splitwindow algorithm, the statistical model is simple and feasible. It does not need to estimate the atmospheric transmissivity and the land surface emissivity. Compar ed with the AWS data, the actual error in 70% retrieved LST is within 4.0℃.