Spectral nudging, a dynamic downscaling method, has been used as a suitable approach to force a regional model to adopt prescribed large-scales over the entire domain, not just at the lateral boundaries, while developing realistic detailed regional features consistent with the large-scales. The aim of this study is to compare a global spectral climate model at high resolution (50 km) and a driven spectral regional climate model over Europe by using the so-called perfect model approach. The spectral nudging method is applied in order to achieve a better representation of large-scale climate over a limited domain. The results show that the regional model driven only at its lateral boundaries presents a summer warm bias in the middle of the domain. This bias disappears when spectral nudging is applied. On the other hand, the smallest scales which are not driven by the spectral nudging are not significantly affected by scale interaction. The only detrimental impact of spectral nudging is a slight precipitation increase in the upper quantiles of precipitation, which can be resolved by large-scale nudging of specific humidity.