In the inverse heat conduction problem (IHCP) of finding the temperature-de
pendent thermophysical properties of materials, the evaluation of the confi
dence region of the unknown functions is at least as important as the numer
ic values of the function itself. The uncertainty of the reconstructed prop
erties arises from various disturbances affecting the measurement instrumen
tation and from an imprecise mathematical description of the experiment. Th
e aim of this work is to consider, besides the effect of random noises supe
rposed onto the signals, the bias arising from the calibration of the senso
rs. By conceiving the calibration process as an integral part of the experi
ment it is possible to quantify the effect of the above bias and to obtain
better estimates and correct confidence regions for the reconstructed prope
rties.