Porous ceramic-metal composites, or cermets, such as nickel zirconia (Ni-YSZ), are widely used as the anode material in solid oxide fuel cells (SOFC). These materials need to enable electrochemical reactions and provide the mechanical support for the layered cell structure. Thus, for the anode supported planar cells, the thermomechanical behavior of the porous cermet directly affects the reliability of the cell. Porous cermets can be viewed as three-phase composites with a random heterogeneous microstructure. While random in nature, the effective properties and overall behavior of such composites can still be linked to specific stochastic functions that describe the microstructure. The main objective of this research was to develop the relationship between the thermomechanical behavior of porous cermets and their random microstructure. The research consists of three components. First, a stochastic reconstruction scheme was developed for the three-phase composite. From this multiple realizations with identical statistical descriptors were constructed for analysis. Secondly, a finite element model was implemented to obtain the effective properties of interest including thermal expansion coefficient, thermal conductivity, and elastic modulus. Lastly, nonlinear material behaviors were investigated, such as damage, plasticity, and creep behavior. It was shown that the computational model linked the statistical features of the microstructure to its overall properties and behavior. Such a predictive computational tool will enable the design of SOFCs with higher reliability and lower costs.