To clarify the spatio-temporal variations of agricultural droughts in the Loess Tableland on the North of the Wei River (LTNWR) from 2001 to 2020, we employed MODIS data to construct Ts-NDVI feature space. Subsequently, the temperature vegetation dryness index (TVDI) was obtained, and the spatio-temporal evolution of droughts was thoroughly examined. Meanwhile, the relationship between droughts and natural factors as well as human activities was analyzed by GeoDetector, and the key driving factors of the spatial differentiation of TVDI were identified. Our findings reveal that 65.04% of the region experienced varying degrees of drought annually, with an average TVDI value over multiple years standing at 0.52, indicating a state of moderate drought across the entire region. Notably, drought conditions were particularly severe during spring. In the study period, TVDI values exhibited fluctuating and declining trends with strong persistence. In turn, approximately 91.38% of regional drought conditions either improved or remained stable during this period. Furthermore, the spatial differentiation of TVDI was pronounced and primarily influenced by factors such as elevation, temperature, population density, and normalized difference vegetation index. The leading factors affecting TVDI varied across different years, reflecting the multifaceted impacts of natural factors and human activities on drought conditions. To address these challenges, integrated measures including climate monitoring, ecological restoration, optimization of agricultural production structures, and improvement of water resources utilization efficiency are essential. These measures will facilitate more accurate prediction and mitigation of the impacts of droughts on regional agricultural production and socio-economic development.