This study investigates climate change impact on hydrological extremes and performs an uncertainty analysis. This work is applied on a French Mediterranean catchment: the Héraut River catchment. With climate change, local water managers need an evaluation of possible future changes in hydro-climatic variables to develop adaptation strategies. This analysis used for the rst time climatic projections derived from a new downscaling method: ADAMONT, with two dierent Radiative Concentration Pathways (RCPs). Hydrological projections are computed with three hydrological models (GR5J, GRSD, TUW) and dierent parameter sets obtained with 29 calibration strategies. The uncertainty analysis is based on the quasi-ergodic analysis of variance (QE-ANOVA), to evaluate the contribution of general circulation models (GCMs), regional circulation models (RCMs), hydrological models (HMs), hydrological parameters, and the internal variability to the total uncertainty. Results for high-ow projections with RCP 4.5 show a signicant increase between 1976-2005 and 2006-2100 (from +5% to +70%), whereas with RCP 8.5 there is no clear trend (from -10% to +60%). These results are signicant and robust for RCP 4.5 but not for RCP 8.5, according to the principle of time of emergence. However, for low-ow projections, results are not signicant, and no clear trend can be drawn. These results are discussed compared to previous hydrological projections on the same catchment, and reveal a higher increase of high-ows over time. Moreover, the hydrological model and calibration strategy contributions to the total uncertainty could be reduced for future investigations. Conclusions of this study could be used for decision makers of the Héraut River catchment to establish adaptation strategies on ood risk.