Modeling of mass transport through nanofiltration membranes has seen an increasing expansion over the last decades in view of proposing predictive tools. Many models were devoted to the prediction of separation by NF and the model proposed here is classically based on the extended Nernst-Planck equation and equilibrium partitioning at the interfaces. The key of such models lies in the estimation of the adjustable parameters. Hence, the feasibility of fully predictive models requires an accurate methodology to identify easily and quickly the input parameters. Among them, the two parameters governing electric and dielectric exclusions have aroused much scientific interest because of their difficult assessment. In this paper, a methodology based on membrane potential (MP) and tangential streaming potential (TSP) measurements is proposed to estimate their values. MP is used to identify the dielectric constant of ternary mixtures which are extrapolated to determine those of salt solutions. In this study, the membrane charge is supposed to vary along the pore length because of the concentration variation. The TSP method is thus implemented to correlate membrane charge and salt concentration via adsorption isotherms, which are introduced in the model to calculate the charge along the pore length. Finally, the ability of the Pore Transport Model (PTM) to predict salt rejections is investigated for two membranes and a satisfactory prediction is obtained. The methodology appears to be convenient for a preliminary estimation of separation performances. (C) 2012 Elsevier B.V. All rights reserved.