Knowledge of the generation of subsurface flow for hillslopes is important for controlling agricultural nonpoint nutrient losses. This study used a physically based hillslope hydrologic model HYDRUS‐2D to assess the sensitivity of simulated subsurface flow to the interactions between precipitation, soil texture and land use. Soil moisture data from 1 January 2013 to 23 August 2014 at two monitoring sites on a tea plantation hillslope were used to calibrate the van Genuchten–Mualem hydraulic parameters for this model. For six different textural classes (loamy sand, loam, silt, silt loam, clay loam and clay) and four land‐use types (tea garden, forest, grassland and bare soil), scenario‐based simulations were carried out for varied precipitation intensities (6.0, 15.0, 30.0, 45.0, 60.0 and 76.0 mm/day) and frequencies (time intervals of 1, 5, 10, 15, 20 and 25 days). Results indicated that the hillslope run‐off was dominated by subsurface flow, which was influenced by precipitation and antecedent moisture conditions. A threshold value of 0.18 m3/m3 of mean hillslope soil moisture was observed for the initiation of subsurface flow. High precipitation intensity (i.e. 75.0 mm/day) substantially increased subsurface flow for all soil textures. In addition, the sensitivity of the bare soil hillslope to rainfall patterns was more than two times higher than that of the vegetated (i.e. grassland, tea garden and forest) hillslope. These findings suggest that extreme precipitation events and land‐use change will increase the risks of subsurface flow on hillslopes. Therefore, optimal fertilizer application strategy and land‐use planning should be proposed for controlling the hillslope nonpoint nutrient losses.