Dynamical downscaling has been recognized as a usefultool not only for the climate community, but also for associatedapplication communities such as the environmental and hydrologicalsocieties. Although climate projection data are available in lowerresolutiongeneral circulation models (GCMs), higher-resolutionclimate projections using regional climate models (RCMs) have beenobtained over various regions of the globe. Various model outputsfrom RCMs with a high resolution of even as high as a few km havebecome available with heavy weight on applications. However, froma scientific point of view in numerical atmospheric modeling, it isnot clear how to objectively judge the degree of added value in theRCM output against the corresponding GCM results. A key factorresponsible for skepticism is based on the fundamental limitations inthe nesting approach between GCMs and RCMs. In this article, wereview the current status of the dynamical downscaling for climateprediction, focusing on basic assumptions that are scrutinized from anumerical weather prediction (NWP) point of view. Uncertainties indownscaling due to the inconsistencies in the physics packagesbetween GCMs and RCMs were revealed. Recommendations on howto tackle the ultimate goal of dynamical downscaling were alsodescribed.