The effects of sample length on the transient measurement results of the water vapor diffusion coefficient of autoclave aerated concrete (AAC) were studied based on the transient technique that solves an inverse problem of one-dimensional mass transfer in a semi-infinite domain. Tests were performed on squarerod-shaped AAC samples having various porosities. The axial sample length was varied from 120 mm to 250 mm while maintaining a constant cross-sectional area of 50 mm x 50 mm. Humidity sensors having two sizes of 6 mm and 12 mm were also adopted, and the transient measurement results were compared with reference values obtained using standardized steady-state method. Results showed that smaller humidity sensors are always preferred and that the appropriate sample length varies with the type of materials. For the specific types of AAC samples studied, the appropriate sample length was found to lie between 150 mm and 200 mm. The use of a shorter (or longer) sample was shown to cause overestimation (or underestimation) of the transient results. It was suggested that the appropriate sample length should be kept short as long as the semi-infinite domain assumption is valid, which could guide the sample length selection for other porous building materials. (C) 2019 Elsevier Ltd. All rights reserved.