We extend the data-based hydrochemical model of Iorgulescu et al. (2005), able to simulate discharge and reactive chemical tracer concentrations (silica and calcium) in streamflow for subcatchments of the Haute-Mentue research basin (Switzerland), to the prediction of additional δ 18O values treated as a conservative tracer. The hydrochemical model is based on a parameterization of three runoff components (direct precipitation (DP), acid soil water (AS), and deep groundwater (GW)) in a chemical mixing model. Each component is modeled through an identical structure consisting of a nonlinear gain and a linear transfer function with two reservoirs (fast/slow) in parallel having a constant partition between them. We formulate a set of hypotheses concerning the isotope characterization of each component to provide additional information about how new rainfall inputs are processed in the hydrological response of the catchment. In particular, the AS component is modeled through a nested structure of hypotheses (models) of increasing complexity. It will be shown that hydrological processes in the hillslope associated with the DP, AS, and GW components are especially effective in filtering of higher-frequency fluctuations in precipitation isotopic ratios at the intraevent, interevent/seasonal, and annual/multiannual timescales. The highly nonlinear and nonstationary AS component represents predominantly "recent" water stored in the upper decimeters of the soil profile. Results also suggest that subsurface pathways are significant for the DP component. A local flow path mechanism is proposed for explaining the large fluxes of subsurface flows.