To address severe grassland degradation, the Chinese government implemented national restoration programmes, which in turn drove a research focus towards assessment of the environmental effectiveness of such initiatives. In this study, net primary productivity (NPP) was used as an indicator for assessing the impacts of land use and cover change (LUCC), improved land-use management and climate change on the grassland ecosystem of the Shiyanghe River Basin. NPP was calculated on the basis of the Carnegie–Ames–Stanford Approach model, which is driven by a Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index and meteorological data. The LUCC data for 2001 and 2009 were derived from MODIS land-cover data. During the study period, the net increase in grassland development was 5105·5 km2, with 80·4% of the newly developed grasslands attributed to desert-to-grassland conversion. The total NPP of grasslands in 2009 increased by 659·62 Gg C compared with that in 2001. The contributions of human activity and climate change to total NPP increase were 133 and −33% respectively. Land conversion and improved management measures directly increased grassland NPP. These factors are dominant positive driving forces, whereas warm and dry climates impose adverse effects on grassland restoration in the study site.