Dynamical downscaling by atmospheric Regional Climate Models (RCMs) forced with low-resolution data should produce fine scale climate details with skill. This is investigated by adopting and extending the Big-Brother approach of Denis et al. (2002). A reference climate is established from a fine resolution RCM simulation in a large domain (the Big-Brother). These Big-Brother (BB) data are degraded by removing small scales, and then used for downscaling by the RCM (the Little Brother) with the same resolution as the BB in three domains of different size. Differences between the Little- and BB are attributed to errors caused by the downscaling. We have furthermore extended the original BB method and investigated the impact of the quality of the driving data. The RCM manage to reproduce the general large scale climate features of the BB when forced with high quality data, but show deficiencies when the driving data differ both in phase and scale from the BB. Forced with data with lower quality on a sufficiently large integration domain and in regions influenced by strong local forcing, the RCM significantly improve the climate statistics for local variables (2 m air-temperature, 10 m wind speed, precipitation). We even found that the improvement increased with domain size.