A methodology has been developed for assessing the sensitivity of electricity and natural gas consumption to climate at regional scales. The approach involves a multiple-regression analysis of historical energy and climate data, and has been applied to eight of the most energy-intensive states, representing 42% of the total annual energy consumption in the United States. Statistical models were developed using two sets of independent variables-primitive variables such as temperature, relative humidity, and wind speed, and derived variables including cooling degree days, heating degree days, and enthalpy latent days. The advantages and disadvantages of both modeling approaches are discussed in this paper, along with sample results for a combined analysis of residential and commercial consumption in eight states.