A segmented dynamic dwelling stock model is proven useful for understanding the development and changes of ageing building stocks, which is highly relevant for renovation measures and estimates of energy use and emissions in aggregated building stocks. In this paper, such a model is developed further for detailed analyses of dwelling stock energy demand and exemplified for the Norwegian dwelling stock 1960-2015. The dwelling stock model simulates the development in stock size and composition and is combined with archetype-specific energy intensities to estimate the total energy demand. After calibrating the model results with statistics, the model is used to explore the phenomena and causes of historical changes. A large-scale improvement of the energy efficiency of the Norwegian dwelling stock has taken place through renovation and construction of new dwellings. A historical shift to more efficient energy carriers and heating systems has had an effect on energy savings in the system, of the same size as the effect of the improved energy efficiency of the stock. However, the total average energy savings per m(2) are offset by changes in user heating habits. A significant decrease in average delivered energy intensity per m(2) is only observed after the introduction of heat pumps. (C) 2016 The Author(s). Published by Elsevier B.V.