In this work, digital image processing and finite element analysis are used to predict physical properties of thermal barrier coatings based on microstructure. Factors that affect thermal conductivity such as porosity, crack morphology, and interface defects are systematically studied. It is shown that transverse cracks can significantly retard the transfer of thermal flux and that the effective TCE and elastic modulus at the coating interface is determined mainly by composition, with interface morphology having little effect. Furthermore, no anisotropy was found at the interface. Unlike traditional property-prediction methods, the methodology presented in this paper reflects real coating microstructures, thus providing more accurate results.