Abstract: Land use types were classified through a combination of visual interpretation and support vector machines, based on 7 Landsat remote sensing images (1974, 1988, 1996, 2001, 2005, 2010 and 2015) of Fuxian Lake Basin in Yunnan Province. The land use dynamic degree, comprehensive degree index, transfer matrix and other models were used to quantitatively analyze the land use types change process in Fuxian Lake Basin. Finally, CA-Markov model was used to predict and simulate future land use. The results are as follows: the area of arbor forest, shrub grass and water area from 1974 to 2015 shows a continuous downward trend, the main direction of diversion is farmland, the increase of construction land area mainly comes from farmland, the trend of wetland and bare land is increasing year by year. The above trends are related to the rapid development of local society and economy. The accuracy of land use is 90.29%, which meets the precision requirement. The trend of land use prediction is arbor forest, water area keeps decreasing trend, shrub grass area, wetland area shows an increasing trend, farmland expansion phenomenon still exists. This study can be provided a reference for the scientific planning, protection, and rational use of land and water resources in the Fuxian Lake Basin.