Vegetation is an important link between water, atmosphere and land, and the growth of vegetation is an important indicator of ecosystem change. Therefore, it is essential to study the dynamic changes of vegetation and predict the vegetation dynamics. Based on the Normalized Difference Vegetation Index (NDVI) and statistical analysis (e.g., trend analysis methods and Hurst exponent), this study investigates the historical dynamic changes of vegetation in China, and the multi-regression model was used to construct a predict model from the perspective of water deficit. The future features were predicted under two representative concentration pathway (RCP4.5 and RCP8.5) scenarios from 12 Coupled Model Inter-comparison Project phase 5 (CMIP5) models. The results showed that 1) most areas (80.1%) of China showed increasing trends in the annual NDVI change during 1982–2015, and the areas showing the degradation trends were mainly found in Northeast China, North Xinjiang and the Qinghai-Tibet Plateau; 2) the prediction model constructed by precipitation and reference crop evapotranspiration (ET₀) can well predict the vegetation dynamics in China; and 3) the future vegetation in China will be better than that in the past, except for some areas (e.g., the northeastern and southeastern parts of China) in spring, and the dynamic changes of vegetation under RCP8.5 scenario will be greater than that under RCP4.5 scenario. Nevertheless, in spring, vegetation degradation cannot be ignored.