Oil well are an important source of geothermal data for studying regional tectonics, reconstructing the evolution of sedimentary basins, and theorizing petroleum generation, migration, and accumulation. A more satisfactory, less laborious method of estimating the thermal conductivity of sedimentary rocks is badly needed. The thermal conductivity of cores from two gas wells in France was measured and correlated with neutron porosity index, sonic interval travel time, bulk density, and gamma-ray logs. To obtain reasonable predictions of conductivity, data were segregated into lithologic groups such as sand-shale, carbonate-shale, and carbonate-sand. A set of regression coefficients in the equation that predicts the conductivity from the logs was calculated for each group. The correlation coefficients between the measured and predicted conductivities of the core samples were 0.81, 0.75, and 0.61, respectively, for the three lithologic groups. The average percent difference between the measured and the calculated conductivities for the lithologies likely to be encountered in practice is 13.4%. We expect this figure can be reduced to 10% by enlarging the data base for calculating the regression coefficients. In the same basin or oil field, the relative errors from well to well probably will be 6% or smaller because the lithology will be nearly homogeneous.