In this study, a comprehensive computational model based on a full statistical approach was developed to investigate the heterogeneous mass transport properties in the metal foam channels, gas diffusion layers (GDLs), and microporous layers (MPLs) of polymer electrolyte fuel cells (PEFCs) at the 95% confidence level. A series of channels, GDLs, and MPLs were, respectively, generated to reflect the random heterogeneous structures and transport characteristics. The critical hydrophobic pore radius in the mixed wettability GDLs was computed by applying a modified Leverett function. Furthermore, the gas transport phenomenon through a sufficient number of porous transport media was simulated using a D3Q19 (ie, three-dimensional, 19 velocities) lattice Boltzmann method, and the corresponding mass transport characteristics were mathematically presented as a function of the porosity. The permeabilities in the channels, GDLs, and MPLs were derived from the pressure gradient and the simulated velocity distribution. It was found that the effective mass diffusion coefficient in the GDLs is mainly influenced by molecular diffusion. Nevertheless, Knudsen diffusion is the dominant mass transfer mechanism in the MPLs, because of small pore diameters. In addition, critical hydrophobic pore radius was derived using a modified Leverett function, which enables to estimate the fraction of pores larger than the critical pore radius in GDLs for effective water transport. Moreover, the interfacial areal contact ratio between two adjacent porous media (ie, channel/GDL and GDL/MPL) was calculated. The calculations indicated that the variation in the local porosity of the porous media has a significant influence on the interfacial connections. The proposed model is expected to improve the prediction performance of porous heterogeneous transport media in electrochemical energy systems and the optimization of porous media structures.