The thermal conductivity has been measured on 25 different samples of rock ranging from a depth of 80 to 900 m in the Echassières borehole (Deep Drilling Project GPF, Massif Central, France). The thermal conductivity of the Bl granite rocks increases from 2.8W/m/C at 180m in depth t0 3.5 W/m/C at 480 m, whilst it is nearly constant in the B2 and B3 granite types at 3.4 W/m/C. These variations are related to the quartz content of the rocks (increase of the quartz content with depth for the Bl type, and constant quartz proportion for the B2 and B3 types). A foliation of the Beauvoir granite with a dip of 40° has been reported from the orientation of micas by Gagny, Jacquot (1986) and by Bernier (1987). This foliation produces no major anisotropy in the thermal conductivity. Additional measurements of quartz orientations have been made using a texture goniometer) showing random quartz crystal orientations. The micaschists of the Sioule series situated at the roof of the Beauvoir granite have a thermal conductivity anisotropy, such that the vertical and horizontal principal conductivities are 2.2 and 3.2 W/m/C respectively. The unusually high mean vertical heat flow of 150 mW/m2 (increasing to.160 mW/m2 at the bottom) calculated using measured temperatures in the borehole is incompatible with the regional mean values reported in the Massif Central, and gives evidence of heat refraction around the borehole resulting from the great thermal conductivity variations from one rock type to another. The shape of the Beauvoir granite has been reconstructed in a 2D vertical section from the preliminary field data. The heat transfer around the Echassières borehole has been determined by numerical simulation and several hypotheses have been tested. The most relevant model involves a mean regional heat flow of 135 mW/m2, a compatible value with some previous measurements reported by Vasseur (1982). From the methodological point of view, this study shows that the interpretation of heat flow measurements made at a regional scale may be significantly affected by heat refraction around high conductivity plutons. A careful interpretation of measured temperatures and heat transfer is necessary in order to obtain reliable regional heat flow densities, as shown by Danis et Royer (1986) in a sedimentary context.