As a dominant ecological stress factor of climate change, soil drought has become a key challenge restricting food security. Based on soil moisture data, this paper uses the cumulative anomaly method, coefficient of variation, Sen + Mann–Kendall trend analysis, and center of gravity shift model to study the spatiotemporal changes in soil drought in Hebei Province from 2001 to 2020 and uses the optimal parameter geographic detector model to analyze the key factors affecting soil drought. The results show the following: (1) over the past 20 years, soil drought in Hebei Province has shown a trend of “first intensifying and then easing”, experiencing two turning points, and its spatial distribution showed significant agglomeration characteristics. (2) Soil moisture showed single-peak seasonal fluctuation, with severe drought from January to May, peak soil moisture from June to August, soil moisture balance from September to October, and soil moisture deficit intensified in winter. (3) Soil moisture stability showed spatial differentiation, being high in the northeast and low in the southwest. Soil drought in about 70% of the region has improved, and the center of gravity of drought-prone areas has moved to the southwest. (4) NDVI and altitude are the main drivers of soil drought spatial differentiation, and the multi-factor interaction shows a nonlinear enhancement effect. Among them, the parameter thresholds such as NDVI > 0.512 and altitude −32~16 m have a significant inhibitory effect on soil drought. This study can make a contribution to improving water resource management and increasing agricultural productivity in the region.