The phenomena of global climate warming and ice-snow melting are intensifying, accelerating ecosystem degradation. Against the backdrop of the "dual carbon" strategy, understanding the response of carbon storage to climate change and human activities has become crucial. This study utilizes annual land use/land cover data from 1990 to 2022 and applies the InVEST model to calculate annual carbon storage. The Sen + MK trend analysis and Spearman correlation analysis were employed to explore the trends and drivers of carbon storage changes in the Qilian Mountains National Park. The results indicate the following: 1) Over the past 33 years, carbon storage exhibited significant spatial heterogeneity, with an average carbon storage of 112.12 t and a slight overall increasing trend. The area with significant changes accounted for only 1.11% of the region; 2) Among different land use/land cover types, forests had the highest carbon storage capacity, whereas grasslands, due to their extensive distribution, had the largest total carbon storage; 3) At a highly significant level (p < 0.001), terrain (excluding aspect), soil, climate, and soil factors were all significantly correlated with carbon storage. In particular, temperature (r = 0.489), evapotranspiration (r = 0.406), and human footprint (r = 0.460) showed significantly stronger positive driving effects on carbon storage compared to precipitation, clay, and fine silt, whereas elevation had a negative impact (r = -0.514). This study reveals the spatiotemporal patterns of carbon storage in the region and deeply explores the key factors influencing carbon storage, providing essential scientific insights for future ecological protection and carbon sink management.