To investigate the possibility of projection of unusual weathers such as hot and cool summer or disaster of heavy rainfall under the global warming condition, we have examined the reconstructed probability of extreme summertime high and low temperatures and local-scale heavy rainfall by using a regional climate model (RCM), Regional Atmospheric Modeling System (RAMS) modified in the Terrestrial Environmental Research Center (TERC). The TERC-RAMS is a nested RCM and has driven by a forcing general circulation field obtained from realistic re-analysis data or Global Climate Models. Preliminary experiments are intended to reconstruct mainly on daily rainfall amount and frequency, and daily maximum and minimum temperatures in Japan. The results showed that the TERC-RAMS could reconstruct the rainfall frequency in one month, but not project the absolute amount of heavy rainfall well. In addition, this model has a systematic bias for under-estimation of temperature. The model experiments have necessity to further development especially on radiation and surface soil/water conditions.