Space heating is the single largest driver of home energy use in England, and internal temperatures are widely used as a proxy for the use of the heating system. To date, the focus has been on estimating average internal temperatures in homes. Little empirical work has been done on analysing daily temperature profiles and what their different shapes and temporal variability can tell us about the physical fabric, heating systems, methods of control, and patterns of occupancy of homes. We analysed a temperature data set, consisting of spot temperature measures taken every 45 minutes for 3 months in 275 dwellings, giving a detailed description of internal temperature in a variety of homes. Building- and socio-demographic variables were collected along with the temperature data. Whilst standard UK domestic energy models make the assumption of a rigid heating pattern and certain internal temperatures, a cluster analysis on the present data showed four distinct profiles of internal temperatures and large variability even within clusters. Translating internal temperatures into statements on the heating system being on or off showed that heating patterns are not fixed within or across homes, i.e. they are probabilistic, not deterministic. We linked the variability in temperatures to socio-demographic variables for which predictions exist (age and household composition), using a regression tree algorithm. The resulting model was run on demographic projections to demonstrate the impact of households aging and shrinking on the average profile of heat demand. Temperature profiles changed only very little from 2008 to 2033. Our analysis suggests the need to revise current assumptions on heating demand temperatures and heating pat-terns, away from a “one size fits all” approach to the integration of variability in temperature and heating de-mand. Linking diversity to explanatory variables would allow targeting interventions to reduce energy demand much more specifically.