The conventional identification of soil water with pre-event water limits deep insights into the involvement of stationary and mobile soil water in subsurface hydrological processes. In three tilled sloping field plots at a hilly area of southwestern China dominated by Entisols, soil water collected with a suction lysimeter was distinguished from the total soil water through an analysis of the stable isotopes deuterium and oxygen-18. Differences in the depth profile of soil water before and after storm events were observed and used to examine how rainwater mixes with soil water and to identify the source contribution of different fractions of soil water in subsurface flow generation. Only water in the 0-10 cm soil layer was significantly affected by evaporation and infiltration. Water in the top 5 cm layer of the soil exhibited the lowest residence time because a storm can replace a substantial proportion of the pre-event water. Soil water at the 10-20 cm depth showed the longest residence time, as indicated by its high proportion of pre-event water. The isotopic signatures demonstrated that piston flow and preferential flow coexisted in this soil. High antecedent soil water content and high rain intensity favor the formation of piston flow. The water collected with the suction lysimeter represented the mobile fraction of the pre-event water in the soil, which effectively participates in the generation of subsurface flow. Newly infiltrated rainwater did not well mix with stationary pre-event water in the soil. The use of recent rainfall to represent mobile soil water may provide a practical solution for overcoming the negative effect of the spatial heterogeneity of the isotopic composition of soil water on hydrograph separation results. Bulk soil water and lysimeter water showed significant differences in isotopic composition under low soil water content or in the top soil layer. Stable isotopes in bulk and lysimeter soil water should be monitored synchronously to reveal the sources and pathways of soil water and their contributions to the generation of subsurface flow in the vadose zone. (C) 2013 Elsevier B.V. All rights reserved.