To assess the distribution characteristics of soil water on the slopes on the Loess Plateau of China, three typical slopes (tableland gully region in Changwu, hilly and gully region in Ansai, and hilly and gully region (water-wind erosion crisscross region) in Shenmu) were selected to analyze the spatial and temporal distribution of soil water and corresponding environmental factors, and the differences in soil water on slopes were compared by geostatistical analysis. According to the mean variation coefficient (CV) along soil profiles at three locations, vertical changes in soil profile at 0–500 cm were divided into four layers: a fast-changing layer (0–40 cm), an active layer (40–100 cm), a sub-active layer (100–200 cm) and a relatively stable layer (200–500 cm). The soil water content had good spatial structure characteristics except for 0–40 cm and 100–200 cm soil layers in the hilly and gully region slope, which can be well simulated by the theoretical semi-variance function model. The surface soil water content of the fitting model was spherical model in general, while the deep soil water content was gaussian model. There were significant differences for the correlations between soil water content in the three slopes and the influencing factors of topography, soil texture and vegetation. There was a significant correlation between soil water content in tableland gully region and the bulk density, altitude and leaf area index (P<0.01). Soil water contents in hilly and gully and water-wind erosion crisscross regions were only significantly correlated with the maximum leaf area and soil texture (P<0.05), respectively. There is a significant difference in soil moisture content between different soil layers in the typical landform types of the Loess Plateau, and geostatistical analysis can be well used to express the spatial heterogeneity of soil water content on a hillslope.