Exergetic analysis of a 2250gpm brackish water desalination plant in California was performed using its operational data. The plant contains nanofiltration, reverse osmosis and electrodialysis reversal units subject to the same source water. The correct definition of exergetic efficiency for such systems was discussed. The effect of feed salinity was also used for further illustrating the difference in the second-law efficiency definitions. The preferred definition can be used to determine the specific energy consumption and makes thermodynamic sense. The nanofiltration, reverse osmosis and electrodialysis reversal units had efficiencies of 0.087%, 0.066% and 0.078%, respectively, which are very low. Various energy recovery devices including a pressure retarded osmosis unit (having infinite area) were applied to the system to see relative increase in second-law efficiency. For the preferred definition, it was seen that the nanofiltration unit had the best efficiency. In terms of alternative designs using energy recovery devices, the pressure retarded osmosis option had efficiencies approximately equal to the hydro-turbine while the highest efficiency occurred through the use of a pressure exchanger at the plant inlet salinity. Therefore, it does not seem to be a practical energy recovery method for the investigated desalination systems.