Grazing and climate change are the two key factors influencing the aboveground net primary productivity (ANPP) in the grassland ecosystems. However, the impacts of grazing on ANPP are still in considerable disagreement, and little is known about its combined effects with climate change in the arid mountainous grassland ecosystems. In this study, we simulated the ANPP dynamics of mountainous grassland ecosystems along altitudinal gradients (i.e. LAG: lower-mountain arid grassland, FMG: forest meadow grassland, and AMG: alpine meadow grassland) over the Northern Tianshan Mountains under the impacts of climate change only (Clm) and by the combined effects with grazing (ClmGra) during 1959-2009 using a modified Biome-BGC model, and modeled the responses of the ANPP to grazing intensities (GIs) by setting 27 different levels of GIs (ranging from 0 to 8 sheep/hm²) as well. Model results indicated that the ANPP increased generally with the warm-humid climate change only in all the three grassland ecosystems, whereas the impacts of grazing on ANPP varied with grassland type. Grazing decreased ANPP in FMG and AMG by 30.0% and 33.2% respectively, in contrast, it resulted in a 1.3% increase of ANPP in LAG because the lower GIs before 1980 generally promoted the ANPP over LAG. Moreover, the responses of ANPP to the increasing GIs varied with grassland type, and the ANPP in LAG decreased significantly after a moderate increase and this changing trend is especially significant during drought years. Nevertheless, the ANPP always decreased non-linearly with increasing GIs in both FMG and AMG. These results suggested that the climate change in the past 50 years may benefit the productivity of the mountainous grassland ecosystem in the arid region of central Asia, but the increasing GIs decreased the productivity substantially. The ANPP of FMG and AMG decreased non-linearly with elevating GIs, but the ANPP may increase under the impact of lower GIs in LAG, especially in drought years. Nevertheless, there are uncertainties in our simulations and the responses of the productivity to increasing GIs together with the underlying mechanisms in arid mountainous grassland ecosystems, more field experiments are needed to validate these preliminary results.