Abstract: With satellite remote sensing data and meteorological data, the surface evapotranspiration in southwest China is estimated by mixed linear dual source remote sensing evapotranspiration model based on surface net radiation, vegetation index, monthly mean temperature and monthly temperature difference and the results are compared with the product data of MODIS. The results show that evapotranspiration estimated by improved remote sensing evapotranspiration model has a good correlation with the MODIS monitoring data, the order of simulation accuracy is autumn > winter > summer > spring; besides, in the past 20 years, the actual evapotranspiration in southwest China has increased obviously, and the evapotranspiration in spring and summer is larger, accounting for 62.3% of the total annual one; in spring, the evapotranspiration shows a descending trend from the southeast to northwest, the actual evapotranspiration in most of Guangxi and southern Yunnan is larger and the west Sichuan is smaller; in summer and autumn, the evapotranspiration shows a decreasing trend from east to west, i.e, from Guangxi, Guizhou, Chongqing to Yunnan and Sichuan; in winter, the trend is decreasing from the south to the north.