The " Energy Golden Triangle" area provides abundant energy for China,and its surface vegetation condition is closely related to the ecological restoration and reconstruction. The dynamic change of vegetation of the border regions of the three provincial areas (Ningxia,Shaanxi and Inner Mongolia) from 2005 to 2015 was studied with the aid of a pixelby- pixel trend analysis method using RS and GIS technology and taking the normalized vegetation index product (MODIS NDVI) as the data source. Geo-weighted regression model is constructed based on a regular grid to explore the effect of 7 factors (elevation,slope,soil clay content,multi-year mean temperature,year average precipitation,distance to coal mine area and distance to road) on vegetation change and its spatial nonstationarity. The results show that the annual average of NDVI of the research area displayed a fluctuating upward trend with a linear tendency being 0.083·(10 a)~(-1)(P< 0.05) during 2005-2015. As for spatial distribution,NDVI showed a decreasing trend from southeast to the northwest. The area with improved NDVI is larger than the degraded area,and accounting for 27.11% and 0.64% of the total study area,respectively. The significant rising area is mainly distributed in the eastern part of Yulin City. NDVI decreased significantly in the whole area,and had stronger spatial clustering (Moran's I is 0.851) from 2013 to 2015,but the degree of variation was spatially heterogeneous,compared with that of 2005 to 2012 years. The change of vegetation in the two periods (2005-2012 and 2012-2015) of the whole region is greatly affected by the natural factors such as climate and so on. The main factors affecting vegetation change are different in time and space. Human activities have double effects on vegetation change.