Vegetation dynamics is crucial for understanding vegetation ecosystem processes in arid and semi-arid regions. The Inner Mongolia Autonomous Region (IMAR) is a typical arid and semi-arid region in China, where vegetation has been significantly altered in response to multiple disturbances over recent decades. However, vegetation dynamics under changing environment and the integrated driving effects on natural and anthropogenic factors are unclear for environmentally sensitive and fragile areas. Therefore, Normalized Difference Vegetation Index (NDVI) as an indicator of vegetation status to systematically analyze the temporal and spatial characteristics of the vegetation dynamics in the IMAR from 2000 to 2020, and we quantify the independent and integrated effects of natural and anthropogenic factors on vegetation changes through Geo-detector. Additionally, we quantitatively separate the driving factors of vegetation from the perspective of dry-wet zones, and explored the different regional vegetation dynamics and its relationship with natural and human activities. The results showed that: (1) Vegetation had generally shown an upward trend with an interannual variability of 0.0186 a−1 from 2000 to 2020. (2) The spatial pattern of vegetation had obvious differences. Most of the improvement was mainly concentrated in the east of IMAR covered 69.14%, and degradation in the western desert region. (3) Natural factors were more influential than anthropogenic factors, precipitation had the greatest explanatory power for the spatial heterogeneity of vegetation with a q value of 80.28%, and the integrated effects on vegetation changes were strongest for precipitation and other drivers. (4) The main drivers affecting NDVI changes are more variable in different wet-dry zones, and precipitation gradients determined explanatory power and the relative importance of natural and anthropogenic factors for vegetation changes. These results contributed more insight into the driving mechanisms underlying vegetation dynamics, while being critical for predicting and evaluating vegetation recovery and vegetation ecosystem stability in the context of global climate change, especially in environmentally sensitive and fragile areas.