One of the challenges in modelling organic solvent nanofiltration (OSN) is to choose and apply a suitable transport model to predict separation performance of a membrane. Although pore-flow models have often been applied to describe membrane performance, it is difficult to obtain some key parameters, including the transport-active pore size distribution, which is a governing characteristic in determining the selectivity. Nanoprobe characterisation is a new imaging technique developed to measure the size of permeation pathways (which we refer to as 'pore size'), and the skin layer thickness in situ. This technique is based on filling the membrane pores with high contrast nanoparticle (NP) imaging agents and then examining the images obtained using transmission electron microscopy (TEM). The objective of this paper is to assess the applicability of a pore-flow model combined with the pore size distribution determined by the nanoprobe imaging technique, to predict polystyrene and NP rejection. Various simulation approaches have been employed. The findings indicate that the simulated rejection using the log-normal probability density function (PDF) to describe the pore size distribution for a range of P84 co-polyimide asymmetric membranes resulted in better rejection predictions for spherical rigid particles than it did for polystyrene solutes. (c) 2013 Elsevier Ltd. All rights reserved.