Pressure retarded osmosis (PRO) was investigated as a viable source of renewable energy. In PRO, water from a low salinity feed solution permeates through a membrane into a pressurized, high salinity draw solution; power is obtained by depressurizing the permeate through a hydroturbine. A PRO model was developed to predict water flux and power density under specific experimental conditions. The model relies on experimental determination of the membrane water permeability coefficient (A), the membrane salt permeability coefficient (B), and the solute resistivity (K). A and B were determined under reverse osmosis conditions, while K was determined under forward osmosis (FO) conditions. The model was tested using experimental results from a bench-scale PRO system. Previous investigations of PRO were unable to verify model predictions due to the lack of suitable membranes and membrane modules. In this investigation, the use of a custom-made laboratory-scale membrane module enabled the collection of experimental PRO data. Results obtained with a flat-sheet cellulose triacetate (CTA) FO membrane and NaCl feed and draw solutions closely matched model predictions. Maximum power densities of 2.7 and 5.1 W/m(2) were observed for 35 and 60 g/L NaCl draw solutions, respectively, at 970 kPa of hydraulic pressure. Power density was substantially reduced due to internal concentration polarization in the asymmetric CTA membranes and, to a lesser degree, to salt passage. External concentration polarization was found to exhibit a relatively small effect on reducing the osmotic pressure driving force. Using the predictive PRO model, optimal membrane characteristics and module configuration can be determined in order to design a system specifically tailored for PRO processes. (C) 2009 Elsevier B.V. All rights reserved.