Probabilistic assessment of hygrothermal performance of building components has received increasing attention by recently published research. Compared to deterministic simulation, probabilistic simulation provides a wide range of possibilities that could cover unexpected scenarios. The inputsapos; uncertainties determine the corresponding range and distribution of simulation outputs. However, the lack of knowledge of input variability becomes one of the obstacles in the risk assessment. In this paper the concept of a stochastic material database for probabilistic building performance simulation was developed and illustrated. The source of uncertainty in the material data was addressed and uncertainty in different data levels was analyzed. In addition to specific materials, generic materials which have the common characteristics of one type of specific material were also included in the database, to take into consideration the situations in which the specific materials of concern are not well known. It is essential that when performing data sampling the correlations between material parameters and the interrelations between material parameters and material functions are taken into account. Probability distributions of material properties in different material categories were analyzed with statistical tests.