In order to quantify the cooling conditions undergone by an ingot during direct-chill (DC) casting, thermocouples were immersed in the liquid pool and consequently entrapped in the solid, thus monitoring the temperature of the metal during its descent. Assuming steady-state thermal conditions, the time-dependent temperatures measured by these thermocouples were then converted into spacedependent temperature profiles. These values were the input of a Maximum A Posteriori (MAP) inverse method described by Rappaz et al. [1] which has been adapted in this case to steady-state thermal conditions. This MAP method permits the deduction of the temperature-dependent thermal conductivity of the alloy, initially, and then of the highly nonuniform heat-flux distribution along the ingot rolling faces, in a second step. The obtained values are in good agreement with literature and clearly reflect the widely different boundary conditions associated with primary cooling (contact with the mold) and secondary cooling (water jet).