In this study, a regional distributed hydrological model is used to perform long-term and flash-flood event simulations, over the Cévennes-Vivarais region (south of France). The objective is to improve our understanding on the role played by geology on the hydrological processes of catchments during two past flash-flood events. This modelling work is based on Vannier et al. ("Regional estimation of catchment-scale soil properties by means of streamflow recession analysis for use in distributed hydrological models", Hydrological Processes, 2014), where streamflow recessions are analysed to estimate the thickness and hydraulic conductivity of weathered rock layers, depending on the geological nature of catchments. Weathered rock layers are thus implemented into the hydrological model CVN-p, and the contribution of these layers is assessed during flash-flood events simulations as well as during inter-event periods. The model is used without any calibration, to test hypotheses on the active hydrological processes. The results point out two different hydrological behaviours, depending on the geology: on crystalline rocks (granite and gneiss), the addition of a weathered rock layer considerably improves the simulated discharges, during flash-flood events as well as during recession periods, and makes the model able to remarkably reproduce the observed streamflow dynamics. For other geologies (schists especially), the benefits are real, but not sufficient to properly simulate the observed streamflow dynamics. These results probably underline the existence of poorly known processes (flow paths, non-linear spilling process) associated with the planar structure of schisty rocks. On a methodological point of view, this study proposes a simple way to account for the additional storage associated with each geological entity, through the addition of a weathered porous rock layer situated below the traditionally-considered upper soil horizons, and shows its applicability and benefits for the simulation of flash flood events at the regional scale.