Erosion is a main form of soil degradation, with severe consequences on slope stability and productivity, and erosion studies are required to predict possible variations of such phenomena, also under climate change scenarios. Here we estimated distributed soil erosion within Valchiavenna valley in the Rhaetian Alps, drained by Mera river, and covering Italy, and Switzerland. We used a Dynamic-RUSLE (D-RUSLE) model, which provides spatially distributed estimates of soil erosion explicitly considering snow dynamic (accumulation/melting) and snow cover, and vegetation seasonality. The model was tuned here during 2010-2019, and validation was pursued using river turbidity data, used to assess riverine sediment transport. The model parameter R-factor for rainfall erosivity was estimated using a hydrological model Poli-Hydro, properly set up in the study area. C-factor for land cover was assessed against land cover maps, with seasonally variable Normalized Difference Vegetation Index from satellite images, to account for variable vegetation stage, and large leaf cover in summer. The K-factor related to erosion susceptibility was evaluated through soil texture and organic content. LS-factor depending on slope was assessed using a DTM. Poli-Hydro and D-RUSLE models were then used to project forward potential soil erosion under climate change scenarios until 2100. Climate series (temperature, precipitation) were generated using 4 shared socio-economic pathways (SSPs) of the Sixth Assessment Report of the IPCC, with 3 global circulation models, properly downscaled locally. We analysed expected soil erosion during 2051-2060, and 2091-2100. We found increase of potential soil erosion, with exception of the EC-Earth model for the SSP2.6. Erosion would especially increase in winter, in response to smaller snow accumulation, and larger liquid rainfall share thereby, and decrease in summer, as due to decreased precipitation. Our results suggest the need for adaptation strategies to counteract increasing soil loss in the future, and may highlight most critical areas of intervention.