In the majority of hygrothermal simulations concerning building materials, moisture transfer is based on Fick’s law, with the assumption of uniform material properties and local equilibrium between liquid and vapor phases. However, the calculated moisture content has been found to change more rapidly than that observed in the vapor absorption process. This study investigated the application of models considering material property heterogeneity and local non-equilibrium, aiming to reproduce the results of moisture absorption and permeation experiments. In a moisture absorption experiment, vapor absorption by a dried aerated concrete specimen was observed. In a moisture permeation experiment, a jar upper was sealed by a specimen, with the relative humidity outside the jar kept constant at a high value, and the humidity change from a low value inside the jar was measured. The results of the moisture absorption experiment showed that the moisture content of the specimen increased gradually over hundreds of days, whereas for the moisture permeation experiment, the relative humidity inside the jar stabilized within approximately a week. Models based on homogeneity and local equilibrium predicted moisture absorption at a rate much faster than the experimental results. After incorporating assumptions of heterogeneity or local non-equilibrium, by identifying suitable values for local resistance to vapor transfer or phase change, the calculation results for both moisture absorption and permeation processes were close to the measured results. This would suggest the necessity of considering local resistance to vapor transfer or phase change to explain the moisture transfer at least for autoclaved aerated concrete.