The Nernst-Planck approach, previously used to model the electrodialytic recovery of uni- or di-valent electrolytes, was employed to comply with the desalination of concentrated brines with an initial NaCl concentration (c(BD0)) ranging from about 1.4 to 3.0 kmol m(-3) and to enucleate the contribution of osmosis and electro-osmosis. The experimental procedure formerly developed was simplified and thus shortened in order to determine just the engineering parameters essential to simulate the desalting process under study. This was further checked by performing a few validation tests in quite different operating conditions from those used in the training tests, that is (i) by changing the electric current (I) step-wisely to simulate the continuous-mode operation of a multistage ED unit, (ii) by reducing the initial salt content of the concentrate by a factor of about 10 and (iii) by spiking the model brine solution with a non-electrolyte osmotic component (i.e., glucose) to enhance water transfer by osmosis. In these tests both membrane resistances were found to be constant and independent of the solute concentration, owing to the fact that the ED desalting process was carried out at electric current densities by far smaller than the limiting one. Either the effective solute (t(B)) or water (t(W)) transport number was independent of I and C(BD0) in the ranges tested. The contribution of solute diffusion was negligible (L(B) approximate to 0), while that of osmosis (L(W)) across the membranes used here was definitively different from zero and had to be taken into account. (C) 2010 Elsevier B.V. All rights reserved.