Comparison of regression analysis of demand each season between residential and commercial areas in Japan is presented. As explanation variables, meteorological parameters i.e. temperature (CDD and HDD) and humidity (RHD), and a dummy for holidays DH are considered influencing the demands. From results, seasonal demand models in residential give slightly higher fitness degree than in commercial area. Among variables used, RHD in CECW for commercial is obtained not significant. However, for meteorological variables, almost all regression coefficients for commercial are larger than for residential area. In forecasting test, models in residential area give higher accuracy than in commercial area.