Over the past few years, anecdotal evidence and controlled studies on predicting the actual energy performance of existing buildings have raised concerns about the accuracy of energy performance simulations using Building Information Models (BIM). Among numerous sources of modeling inaccuracy, the wrong thermal properties of solid assemblies can significantly impact the reliability of the BIM-based energy analysis. Without considering the diminishing thermal resistance (i.e., R-value) of building materials caused by their deterioration and updating such material properties in BIM, BIM-based energy simulation of existing buildings could be unreliable for retrofit decision-making purposes. This paper proposes an automated method for updating the R-value of building elements in BIM. By exploring 3D spatio-thermal models reconstructed from 2D thermal images, the actual R-values are calculated at the level of 3D points. The calculated actual R-values are used to automatically update the thermal properties of the corresponding gbXML-based BIM elements. For validation, several experiments on measuring and updating the R-value of BIM elements are conducted in an existing residential building. Using the proposed method, BIM-based energy analysis tool scan more accurately model the as-is conditions of existing buildings and estimate their actual energy performances.