Desertification in the arid and semiarid regions directly influences the density and growth status of vegetation, NDVI (Normalized Difference Vegetation Index) has been widely used to monitor vegetation changes. This study analyzed the spatial patters of vegetation activity and its temporal variability in Tarim Basin, Xinjiang, China since 1998 to 2007 with NDVI data derived from SPOT4 Vegetation. The coefficient of variation (CoV) of the NDVI was used as a parameter to characterize the change of vegetation and to compare the amount of variation in different sets of sample data. The method of quantifying changes in CoV values for each pixel was based on linear regression. The slope of linear regression was acted as the criterion for the change direction: pixels with a negative slope are considered to represent ground area with decreasing amounts of vegetation, vice versa. In this paper, We calculated (1) the inter-annual CoV based on the yearly ∑NDVI, the sum of the monthly NDVI in the growing season (from April to October), for each pixel between 1998-2007 to reveal the spatial patterns of vegetation activity, (2) the intra-annual CoV based on monthly NDVI by MVC to reflect vegetation seasonal dynamics, (3) the slope (κ) of the intra-annual CoV regression line for each pixel to identify the overall long-term trend of vegetation dynamics. This experiment demonstrated the feasibility of applying the CoV and its regression analysis based on long term SPOT-VGT NDVI time-series data for vegetation dynamics monitoring.