Measurements of the diffusive permeances of water, NaCl, and ethanol through several, unoptimized membranes are presented. Such data can facilitate analysis and development of water recovery from highly impaired sources using hybrid processes based on forward (direct) osmosis (FO) with aqueous ethanol solutions as the "osmotic" agent. The membranes we have studied include anion and cation exchange materials and cross-linked poly(vinylalcohol) (PVA) gels, the latter being a membrane chemistry commercially used for ethanol dehydration via pervaporation. The measured transport properties are reported and suitability of these materials for an FO-based water recovery process is discussed in the context of process simulations. A major economic consideration for most FO processes is the cost of the lost "osmotic" agent. Thus, we focus our evaluations around the relative selectivity of the membranes for water versus ethanol, alpha(w/E). We made measurements using initial NaCl concentrations of 3.5-9.5 mass% with constant ethanol mole fraction (x(ErOH) = 0.2) in the draw solution. We also varied ethanol mole fraction with a constant initial salt concentration (3.5 mass%). The water/ethanol selectivity of all the membranes showed variability with these changing solution conditions. The average alpha(w/E) in PVA, and Selemion AMV and CMV (anion and cation exchange membranes, respectively), over the range of salt concentrations, were similar to 7, 11, and 34, respectively. However, all the membranes exhibited a high degree of variability with respect to the changing boundary conditions (interfacial compositions) over the course of these measurements, likely due to differential swelling. We concluded that selectivities reported for pervaporation-based separations do not necessarily translate to dialysis-based ones, and that significantly higher selectivities than we obtained are likely required for commercially viable processes. (c) 2008 Elsevier B.V. All rights reserved.