[Objective] The response relationships of the net primary productivity loss （ΔNPP） of different vegetation types caused by drought of different vegetation types to various precipitation annual patterns were evaluated， in order to gain an in-depth understanding about the responses of vegetation to climate extreme events and provide a theoretical basis for regional drought prevention. [Methods] Drought characteristics was quantified by standardized precipitation evapotranspiration index （SPEI）， and NPP was estimated based on MODIS remote sensing data by carnegie-ames-stanford approach （CASA） model and annual SPEI during 2001 to 2020 in the middle arid zone of Ningxia Hui Autonomous Region was calculated to explore the differences of net primary productivity（NPP） of different vegetation types to SPEI， and the correlation between NPP loss （ΔNPP） due to drought and SPEI index by using the methods of trend analysis， Mann-Kendall test and correlation analysis. [Results] ① From 2001 to 2020， the SPEI index indicated that the drought trend in the study area was generally slowing down， and drought was still dominant in spatial dimension， but there was a trend of humidification. ② ΔNPP in the study area showed a totally decreasing trend of fluctuation； spatially， the ΔNPP in the eastern part of the study area decreased significantly. ③ The correlation analysis between SPEI index and ΔNPP showed that the negative correlation was dominant， but there were obvious spatial-temporal differences due to different drought grades. ④ The response of ΔNPP of different vegetation types to annual SPEI was different， under severe and extreme drought stresses， the NPP loss of forest was the largest， followed by crop， and shrubs was the least. [Conclusion] At present， the middle arid zone of Ningxia shows a trend of humidification. In order to improve the resistance of vegetation to drought in the study area， the complexity and diversity of vegetation structure and composition should be improved.