Tourist attractions are not only the prerequisites and core carriers of tourism development, but also the most important and fundamental tourism supply. Based on the data of high- level tourist attractions in Western Hunan, this paper introduces the index of development difference of high-level scenic spots, and uses spatial statistical methods such as Standard Deviation Ellipse, nuclear density estimation and global spatial autocorrelation to measure the spatio- temporal distribution of high- level tourist attractions in 2007, 2012 and 2016. And then, it uses traditional ordinary least squares (OLS) model and geographic weighted regression (GWR) model to analyze the spatial heterogeneity of the influencing factors in the high- level tourist attractions of Western Hunan. Results show that: Generally, the spatial distribution of high-level tourist attractions in Western Hunan has a NE-SW trending, and the overall trend is a small counter- clockwise rotation; the spatial pattern of high- level tourist attractions shows the evolution from "three cores and two districts" to "single pole and many points" and then to "blossom all over" driven by dual-core. The spatial positive correlation of the number of high- level tourist attractions is decreasing, and its spatial distribution shows a certain dependence; regional economic development level, government policy support, traffic conditions and tourism resources have significant influence on the spatial distribution of the high-level tourist attractions in Western Hunan, and different influencing factors show a certain spatial difference. The research conclusion provides scientific basis and reference value for deepening the reform of mechanism of tourist attractions in the context of all-for-one tourism. [ABSTRACT FROM AUTHOR]