Study region: The Jing River Basin (JRB) and Tao River Basin (TRB) on the Loess Plateau (LP), China. Study focus: Baseflow plays a critical role in hydrological drought evolution, yet baseflow drought and its propagation processes remain poorly understood. Moreover, hydrological drought assessments are often constrained by low simulation accuracy and considerable uncertainty. To address these limitations, this study applies an improved Soil and Water Assessment Tool (SWAT) model in combination with the Eckhardt digital filter and run theory to simulate streamflow, separate baseflow, and identify drought events. The characteristics of meteorological, streamflow, and baseflow droughts, together with their propagation dynamics, are systematically analyzed and compared. New hydrological insights for the region: The improved SWAT significantly enhances streamflow simulation accuracy, providing a reliable basis for drought analysis. Meteorological droughts occur more frequently than hydrological droughts, whereas hydrological droughts, particularly baseflow droughts, exhibit longer duration and greater severity. Propagation from meteorological drought to streamflow drought shows higher effective propagation rates and shorter response times than propagation to baseflow drought, indicating a more sensitive response of streamflow to precipitation variability. Basin comparison reveals faster propagation to streamflow drought in the TRB than in the JRB, while propagation to baseflow drought is slower and shows contrasting temporal patterns. These findings highlight the importance of baseflow drought in hydrological drought analysis and support watershed scale water resource management.