A critical parameter for evaluating the resistance of concrete against chloride diffusion is effective diffusivity. While it can be obtained experimentally, this paper concerns computer simulations to estimate diffusivity of mortars as influenced by paste, aggregate, and interfacial effects. However, when performing such simulations, the size of the mortar sample modeled must be large enough to reliably characterize these randomly distributed effects, while remaining small enough for computational efficiency. The purpose of this paper is to report the influence of sample size on computed results and to recommend the smallest specimen size needed to obtain reliable estimates of diffusivity. (C) 2011 Elsevier Ltd. All rights reserved.