The Budyko framework has been successfully used to evaluate variations in hydrological processes and their responses to climate change and human activity. Recently, this framework was extended to estimate the mean annual terrestrial gross primary productivity (GPP) by defining two GPP potentials and a control parameter (n). However, the feasibility of the GPP-Budyko framework at the growing season scale has not been validated and an attribution analysis of GPP changes under this framework has not been conducted. Therefore, focusing on 18 inland river basins of Northwestern China (IRB-NWC), the GPP-Budyko framework was improved by considering basin water storage change, and its adaptability was validated at the growing season scale. Furthermore, the impacts of climate change and human activity on the growing season GPP were quantified using an elasticity analysis approach. The results showed that the improved Budyko framework effectively captured the effect of vital climatic constraints on the growing season GPP. The control parameter n increased in most basins from 1982 to 2018. Additionally, the mean growing season GPP was 88.46 gC m- 2 with an increasing rate of 1.36 gC m- 2 a-1. The increasing growing season GPP was dominated by human activities, with a relative contribution rate of 59.84 %, ranging from 37.54 % to 89.21 %. The climate change was also an important factor affecting the growing season GPP with a median relative contribution of 40.16 %, ranging from 10.79 % to 62.46 %. This study provided a practical tool for estimating GPP at the growing season scale and offered opportunities to evaluate the effects of ecological projects and management on vegetation restoration.