Abstract In the debate on predicting household energy savings, the temperature takeback – an increased indoor temperature after an energy efficient retrofit – is often blamed for offsetting part of the potential energy savings. Mostly, it is attributed to inhabitants grading up their heating behaviour to the lower energy cost after retrofit. However, even if inhabitants do not change their heating pattern, the indoor temperature will still rise after retrofit due to physical processes: warmer unheated zones and less temperature drop between two heating periods. This paper uses building energy simulation tools to assess the extent of these physical processes in the overall temperature rise. An existing terraced house is modelled and fictitious renovation measures are imposed, keeping the heating patterns unchanged. For the case analysed, a heating season mean indoor temperature rise of about 1 °C is found, being in the same order of magnitude as empirically detected temperature changes. This suggests that the remaining behavioural aspect of the temperature takeback might be smaller than generally assumed. In addition, the comparison is made with a calculation method based on the EPBD regulation that does not take into account the physical temperature rise. The latter method overestimates the potential energy savings by about 6%.