[Objective] This study examines the impact of climate change and human activities on baseflow variability across four representative watersheds within the Loess Plateau: the Wuding River, Kuye River, Yan River, and Tutai River. [Methods] Utilizing daily flow data from 1965 to 2017 for the Wuding, Kuye, and Yan rivers, and from 1983 to 2017 for the Tutai River, alongside multi-site meteorological data, we evaluated the applicability of nine baseflow separation methods and characterized baseflow variations. Contributions of climate change and human activities to baseflow changes were quantified using the elasticity coefficient method within the Budyko framework. [Results] The F4 digital filtering method exhibited superior performance in terms of stability, dispersion, and error metrics, making it the most suitable method for the typical watersheds of the Loess Plateau. Trends in annual baseflow depth, annual runoff depth, and annual baseflow index were consistent across the four watersheds, with multi-year average baseflow indices (BFI) of 0.64, 0.42, 0.28, and 0.49, respectively. Significant decreasing trends (p<0.01) were observed in both annual baseflow depth and runoff depth, while the annual baseflow index showed increasing trends. The contributions of precipitation, potential evapotranspiration, and human activities to the changes in baseflow ranged from -56.72% to 3.31%, -0.81% to 55.16%, and 10.92% to 104.37%, respectively. [Conclusion] The findings underscore that human activities are the principal drivers influencing baseflow evolution in the Loess Plateau area, offering theoretical support for integrated water resource management in the region.