Cosmic-ray neutron sensing (CRNS) can monitor soil water content (SWC) in the 100-meter scale source area, but its applicability and accuracy in the loess hilly-gully region with complex terrain require validation. In this study, a typical sub-watershed in loess hilly-gully region was selected as the research object, and CRNS was used as the technical means. Twelve large-scale distributed oven-drying method measurements (0-350 m) were carried out in 2018-2019. Firstly, we analyzed the characteristics of soil moisture changes in the whole study area and gully region during the study period; then , according to the soil moisture content obtained by three weighting methods (dynamic weighting, average and uniform weighting) was calibrated to the N₀ value to assess the measurement footprint and accuracy of the CRNS method in the loess hilly -gully region; and we analyzed the stability of the continuous CRNS-inversed soil water content (SWC) and its responsiveness to rainfall events by comparing them with the continuous monitoring data of the time-domain reflectance (TDR) method. The results showed that the average SWC of the whole study area measured by the oven-drying method was smaller than the average SWC of the gully region, and both of them have the same trend of change; the average values of the horizontal footprint radius and the vertical footprint of the CRNS method were 336.0 m and 23.4 cm, respectively, and the horizontal footprint varied steadily under the influence of air pressure, while the vertical footprint showed obvious fluctuation according to the distance from the center of the instrument and the soil moisture condition; the R^₂ of the linear fit between the CRNS method and the oven-drying method was greater than 0.9, and the RMSE was in the range of 0.027~0.037 cm₃^/cm₃ , which indicated that the CRNS method was able to accurately measure the soil moisture content of the region in the loess hilly-gully region. In terms of stability, the trends of regional CRNS-inversed SWC and the average soil water content at 20 cm depth in the gully region obtained by the TDR method were identical except for the winter freezing period, and both of them responded significantly to the precipitation process, and the sensitivity of the CRNS method was better. In conclusion, the CRNS method, with its large-scale coverage, non-destructiveness and adaptability to loess texture and topography, can accurately and reliably measure the average SWC at the hectometer scale in the loess hilly-gully region, and can make up for the deficiencies of the traditional point-scale and remote sensing methods.