One of the strategies for decarbonizing the energy mix is to increase the use of district heat networks, for distribution of waste heat or heat produced by renewable energy sources. However, planning such networks needs a geo-dependent energy database concerning demand and supply, incorporating their dependency on space and time. The present paper concerns the development of a bottom up statistical extrapolation model for estimating the heat demand of the Swiss building stock. The database is constructed on top of the Swiss building register, which contains basic data on building category, age, ground area and number of floors, as well as area devoted to dwellings. The model itself concerns: (i) estimation of the heated area of each building; (ii) estimation of the heat demand of each building, with disaggregation in terms of space heating (SH) and domestic hot water (DHW). The model is calibrated by way of recorded data concerning actual yearly energy consumption of around 27'000 buildings, with a classification by building category and age. The disaggregation in terms of SH and DHW allows for climatic correction of the observed data within a common climate basis. In a second step this calibration data is used for extrapolation of the demand on the entire Swiss building stock, for which the SH component is corrected by taking into account local climate characteristics. Finally, statistical methods tend to quantify the uncertainty inherent to the use of average demand values by age and category, in relation to the level of spatial aggregation.