Random Uncertainty is the inherent attribute of natural and engineering materials; the heat conduction behavior is one of the most important questions in servicing process of thermal protection system, and the research on the heat conduction problem of random materials can provide theoretical basis for integrative design of material and structure in thermal protection system. A novel statistical multiscale analysis method based on two-scale asymptotic expansions is proposed to predict heat conduction performances of random porous materials. The statistical multiscale formulations and statistical multiscale boundary element algorithm are brought forward. Besides, the validity of the proposed method by comparison with theoretical results was verified. Finally, the effect of microstructures on the macroscopic thermal properties for the porous ceramic materials is investigated. Numerical results prove the accuracy and efficiency of our method for multiscale simulation of heat conduction problem in random porous materials.