The logarithmic transformation of the rear face temperature rise issued from a flash experiment can be interpreted as adding up a heat loss and radial effect component to the one for the adiabatic model. Two data reduction methods based on this transformation are proposed with the common main goal to identify one parameter depending upon the sample’s thickness and thermal diffusivity, not upon the other experimental conditions. They allow one to determine thermal diffusivity with good accuracy for nonuniform pulses and high Blot numbers.