High-resolution downscaling is vital to project climate extremes and their future changes by resolving fine topography reasonably well, which is a key to represent local climatology and impacts of weather extremes. A direct dynamical downscaling with a regional climate model (RCM) embedded within an atmosphere-ocean coupled general circulation model (AOGCM) is commonly used but is subject to systematic biases in their present-day simulations of AOGCM, which may cause unexpected effects on future projections and lead to difficult interpretation of climate change. In a high-resolution atmospheric general circulation model (AGCM)-RCM system, the present-day climate in AGCM is forced by observed sea surface temperature (SST) and sea-ice distribution. Then, the future climate is calculated with the "future" boundary conditions (SST and sea-ice), which are created by adding their future changes projected by AOGCM to the observed present-day values, besides the future radiative forcing. This system is one of methods to minimize the effects of such biases. A Meteorological Research Institute AGCM with 20-km grids is successfully applied to project future changes in weather extremes such as tropical cyclones and rain systems that cause heavy rainfall and strong winds. Regional downscaling with 5-km mesh RCM is then performed over certain area to investigate local extreme rainfall events and their future changes. In this paper, we review various downscaling methods and try to rationalize a use of high-resolution AGCM-RCM system. Keywords regional climate information; dynamical downscaling; GCM; RCM Corresponding author: Akio Kitoh, Faculty of Life and Environmental Sciences, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8572, Japan E-mail: kito.akio.ff@u.tsukuba.ac.jp ©2016, Meteorological Society of Japan Journal of the Meteorological Society of Japan Vol. 94A 2 tical downscaling can provide climate information at the smaller scales. Use of high-resolution dynamical models can add values through better representation of topography and coastlines, and also from improved representation of small-scale processes like con vective precipitation.Dynamical downscaling methods include nested regional climate models (RCMs) and high-resolution atmosphere-only general circulation models (AGCMs). Here, we discuss both methods focusing on the latter approach. For the short-and medium-range (up to 2 weeks) numerical weather predictions (NWPs), world centers are now using AGCMs with grid size around 15-30 km. Some NWP centers use a regional model with a few km grids for veryshort-range forecasts. Due to increasing availability of high-performance computers, we are now capable to modify this NWP-type downscaling system for use in climate change projections. In Section 2, various dynamical downscaling methods are discussed. Section 3 shows the global high-resolution Meteorological research Institute (MRI)-AGCM, and Section 4 shows current projects for further dynamical downscaling by the non-hydrostatic regional clim...