Within the vast category of unconventional resources, heavy oils play an essential role as related resources are abundant throughout the world and the amount of oil produced using thermal methods is significant. Simulators for thermo–hydro–dynamic modeling, as a mandatory tool in oilfield development, are continuously improving. However, the present paper shows that software capabilities for the integration of data on the rock thermal properties necessary for modeling are limited, outdated in some aspects, and require revision. In this paper, it is demonstrated that a characteristic lack of reliable data on rock thermal properties also leads to significant errors in the parameters characterizing oil recovery efficiency. A set of advanced methods and equipment for obtaining reliable data on thermal properties is presented, and a new, vast set of experimental data on formation thermal properties obtained from the Karabikulovskoye heavy oil field (Russia) is described. The time-dependent results of modeling oil recovery at the field segment using the steam-assisted gravity drainage method with both published and new data are discussed. It is shown that the lack of experimental data leads to significant errors in the evaluation of the cumulative oil production (up to 20%) and the cumulative steam/oil ratio (up to 52%).