By area-wide implementation of smart metering, large amounts of individual electricity consumption data with a high temporal resolution become available. We use multiple regression models for hourly electricity consumption in Norwegian dwellings, based on panel data consisting of hourly smart meter data, weather data, and response data from a household survey. Two models based on daily and hourly mean values of outdoor temperature, respectively, are compared and discussed. Our results indicate that daily mean outdoor temperature - represented by heating degree day - can serve as weather related input variable for modeling aggregate hourly electricity consumption. The regression models are further used to break down hourly electricity consumption into two components, representing modeled consumption for space heating and other electric appliances, respectively. Thus, without submetering electric heating equipment an estimate for heating energy consumption is available, and can be used for evaluating different demand side management options, e.g. fuel substitution or load control. Moreover, the models can be used for forecasting aggregate regional electricity consumption in the Norwegian household sector with a high temporal resolution, as e.g. changes in regional climatic conditions, dwelling structure, and demographic factors can be taken into account. (C) 2016 Elsevier B.V. All rights reserved.