Geothermal heat flux under the Antarctic ice is one of the least known parameters. Different methods (based on e.g., magnetic or seismic data) have been applied in recent years to quantify the thermal structure and the geothermal heat flux, resulting in vastly different estimates. In this study, we use a Bayesian Monte-Carlo-Markov-Chain approach to explore the consistency of such models and to which degree lateral variations of the thermal parameters are required. Hereby, we evaluate the input from different lithospheric models and how they influence surface heat flux. We demonstrate that both Curie isotherm and heat production are dominating parameters for the thermal calculation and that use of incorrect models or sparsely available data lead to unreliable results. As an alternative approach, geological information should be coupled with geophysical data analysis, as we demonstrate for the Antarctic Peninsula.