Water use efficiency (WUE) is the regulation index of leaf photosynthesis on the water physiological process, as the key to link the relationship between the ecosystem carbon cycle and the water cycle, and directly reflects the rapid adjustment of the ecosystem to the site environment and the adaptation strategy of resource changes. The Gross primary production (GPP) temporal and spatial characteristics derived from the data of MODIS-NDVI by the Light Energy Utilization Model, and the Evapotranspiration (ET) temporal and spatial characteristics derived from the meteorological data by the Thornthwaite Evapotranspiration Model during 2000-2019 in Shiyang River Basin (SRB). In this research, we estimated the spatial pattern and temporal variations of the GPP, ET and WUE in different vegetation types from 2000 to 2019, investigated the co-relationship between GPP/ET/WUE and Vapor pressure deficit (VPD), and discussed the response of adaptation strategies for water utilization and stress in the arid area among the different vegetation. The results show that: (1) from 2000 to 2019, the average annual value of total WUE, GPP, and ET in SRB were 0.80 gC m^-2 mm^-1, 256.52 gC/m², and 302.52 mm, respectively; The spatial patterns of average annual WUE, GPP and ET in the SRB's vegetated areas varied widely, and the values decreased from upstream on the south side of the Qilian Mountains and middle to downstream on the north side of the desert. (2) During the 20 years, the average slope of WUE, GPP, and ET showed an obvious upward trend in almost whole basin, with the rate of 0.017 gC m^-2 mm^-1 a^-1, 6.99 gC m^-2 a^-1 and 3.80 mm/a. The change rates of WUE, GPP and ET within the basin exhibited an upward trend, except for Liangzhou and the areas surrounding Minqin cropland according to the urbanization and industrial adjustment. (3) The WUE, GPP and ET annual average values, and slope of trends among the different land use types also varied significantly. Among various vegetation types, the highest values of WUE, GPP and ET appeared in forests, and the lowest values presented in wetlands; the highest change rate of different vegetation was forest, and the lowest one was the wetland. (4) Correlation analysis and statistical results indicated that the correlation distribution characteristics of GPP/ET/WUE and VPD in the study area were highly consistent with the runoff direction of the SRB. As the WUE was determined by GPP and ET, the correlation distribution between WUE and VPD had obviously spatial heterogeneity with the land-use and SRB runoff direction. The positive correlation was distributed in the eastern part of the Hexi Corridor oasis and the eastern desert area, and the negative correlation was distributed in the high-altitude areas of the Qilian Mountains, the middle of the Hexi Corridor oasis and the western part of the desert area. Overall, correctly revealing the spatial differentiation and changing trend of GPP/ET/WUE of inland river vegetation and its response to the driving factor VPD was the premise of understanding the adaptive strategy of vegetation in arid areas at the watershed scale to the intensification of global drought, to provide decision-making services for ecological protection.