Vegetation degradation is one of the key subjects in the study of global environmental change, and the Normalized Difference Vegetation Index (NDVI) is generally recognized as a good indicator of terrestrial vegetation productivity and growth status. To evaluate the vegetation dynamic changes in the Jinghe watershed on Loess Plateau from 1982 to 2003, major methods of changing slope, principal component analysis and correlation analysis are employed with 8km resolution NOAA-NDVI time series data. Based on these analyses, the relationship between precipitation and NDVI is discussed. Results show that there is little change in both the amplitude and variety of NDVI during the past 22 years. Vegetation in the upper stream areas, typically the watershed marginal mountain areas, changes significantly. A trend analysis shows the similar finding on that vegetation dynamics in different areas tends to be induced by climate change and human land use transformation. A standardized principal components analysis indicates that the first two components, or PC1 and PC2, are closely related to vegetation and climate changes. While PC3 and PC4 are connected with floodwater in flooding seasons, PC5 and PC6 reflect the effects of human activities. Finally, the correlation analysis between NDVI and precipitation shows that they have a close positive relationship in this region. The rainfall sensitivity threshold reaches 550mm or even higher.