We conducted comparative measurements of thermal properties of samples from nine cores of the ICDP COSC-1 borehole and four widely used rock references, using a steady-state and a transient divided-bar device, a transient plane source device, a modified Ångström device, as well as two optical thermal conductivity scanners. In addition, a caloric method provided benchmark values for specific heat capacity. A complementary thin-section analysis of the COSC-1 samples allowed us to calculate specific heat capacity according to Kopp’s law and thermal conductivity according to commonly used mixing models. Our results demonstrate agreement between the various test methods within |$\pm 10$| per cent for about one half of the investigated samples. Furthermore, almost all results for specific heat capacity agree with the predictions of Kopp’s law, though the significance of this correspondence is limited owing to large uncertainties in the experimental and theoretical values. The results for thermal conductivity fall within the most extreme theoretical bounds that account for anisotropy but for an amphibolite. Thermal anisotropy seems to contribute significantly to the deviations between results of the different transient methods that, however, cannot be reconciled by the available theoretical relations for apparent thermal conductivity of transversely isotropic materials. The combination of characteristic investigation volume of the individual methods and sample heterogeneity has to be considered responsible for variability of results, too, an issue whose clarification is calling for dedicated numerical modelling in the future, with the prospect to characterize thermal heterogeneity from observed differences.