This dataset provides high-resolution (500 m × 500 m) gridded maps of soil hydraulic properties (SHPs) across China’s drylands, derived from field-measured data and machine learning-based pedo-transfer functions (PTFs). The dataset includes spatially explicit predictions of five primary SHPs—saturated hydraulic conductivity (cm/min), saturated water content (cm³/cm³), residual water content (cm³/cm³), and the van Genuchten parameters α (1/cm) and n (-)—at six standard soil depths (0–5, 5–10, 25–30, 55–60, 95–100, and 195–200 cm), covering the full 0–200 cm soil profile. Additionally, four derived SHPs—field capacity (cm³/cm³), wilting point (cm³/cm³), plant available water (cm³/cm³), and soil macroporosity—were calculated from the primary SHPs and provided at the same resolution and depths. The gridded dataset captures the spatial variability and vertical heterogeneity of SHPs across diverse soil types, land uses, and climatic zones in China’s drylands. This dataset supports improved parameterization in land surface and earth system models and facilitates more accurate simulations of hydrological and ecological processes in arid and semi-arid regions. Data are provided in NetCDF format, organized by soil variable, and are suitable for direct integration into regional to continental-scale modeling frameworks.