This data set provides a set of residential flood loss maps (ESRI Shapefiles) for the German part of the Danube catchment for current and future climate based on a stochastic event set of flood hazard footprints (Schröter et al. 2017; http://doi.org/10.5880/GFZ.5.4.2017.003). The multi-polygon maps provide flood loss in EUR for residential land use areas according to the ATKIS (Authoritative Topographic Cartographic Information System) codes residential areas (2111) and areas of mixed use (2113), (BKG GEODATENZENTRUM: ATKIS-Basis-DLM, 2005). Loss values are calculated using the FloodLossEstimationMOdel for the residential sector (FLEMOps+r) developed by Elmer et al. (2010) in combination with exposure data based on total replacement costs for residential buildings (Kleist et al., 2006). Asset values with a spatial resolution corresponding to the underlying inundation depth maps of the stochastic event set (100x100 m) have been derived by applying a binary disaggregation method and using the digital basic landscape model ATKIS as ancillary information (Wünsch et al. 2009). The flood event sets are derived for the historical period (1970-1990) and two RCPs (4.5 and 8.5) for the near future (2020-2049) and far future (2070-2099) for four CORDEX models. These flood event sets are created within continuous long-term simulations of a coupled model chain including the IMAGE stochastic multi-variable, multi-site weather generator, the eco-hydrological model SWIM and 1D river network coupled with 2D hydro-numeric hinterland inundation model, see Schröter et al. (2017) for further details The data have been produced within the OASIS+ demonstrator project 'Future Danube Multi Hazard and Risk Model' funded by Climate-KIC in the period from January 2016 to December 2017. Key features:• Flood loss maps for residential areas in the German part of the Danube catchment from stochastic flood event sets for current and future climate.• High spatial resolution for ATKIS residential land use areas intersected with 100x100 m inundation depth maps.• Flood loss scenarios for historical period (1970-1990) and two RCPs (4.5 and 8.5) for the near future (2020-2049) and far future (2070-2099) from four CORDEX models Key usage:• Large-scale flood risk assessment• Future flood risk assessment• Flood risk management with long-term perspective A full description of the data provenance and specification is given in the README_Schroeter-et-al-2017-004.txt file available in the data download section at this DOI Landing Page.