The porosity of cement paste affects its mechanical and thermal properties. Even when two specimens have the same degree of porosity each other, the void distribution considerably affects the behavior of material. To evaluate the material properties of cement pastes statistically, a sensitivity analysis using a First-Order Second-Moment (FOSM) method can be used. This is a probabilistic method to determine the probability distribution of output variables with random input variables. The porosity(φ) and continuity of void (L p area, Ω) are selected as input variables, and the thermal conductivity and stiffness of cement paste are selected as output variables. When a virtual specimen is generated from micro-level computerized tomographic (μ-CT) images of a real cement paste specimen, the specimens that have objective microstructures can be obtained using a reconstruction process. In this study, statistical distributions of input variables are from 64 virtual specimens and output variables are estimated from reconstructed specimens using finite element analysis. Based on sensitivity analysis, sensitivity measures of material properties on both characterizations are evaluated. From this results, the probability distributions of the responses can be estimated and the relation between input and output variables can be evaluated.