Exergy analysis of micro-organic Rankine heat engines is performed to identify the most suitable engine for driving a small scale reverse osmosis desalination system. Three modified engines derived from simple Rankine engine using regeneration (incorporation of regenerator or feedliquid heaters) are analyzed through a novel approach, called exergy-topological method based on the combination of exergy flow graphs, exergy loss graphs, and therrnoeconomic graphs. For the investigations, three working fluids are considered: R134a, R245fa and R600. The incorporated devices produce different results with different fluids. Exergy destruction throughout the systems operating with R134a was quantified and illustrated using exergy diagrams. The sites with greater exergy destruction include turbine, evaporator and feedliquid heaters. The most critical components include evaporator, turbine and mixing units. A regenerative heat exchanger has positive effects only when the engine operates with dry fluids; feedliquid heaters improve the degree of thermodynamic perfection of the system but lead to loss in exergetic efficiency. Although, different modifications produce better energy conversion and less exergy destroyed, the improvements are not significant enough and subsequent modifications of the simple Rankine engine cannot be considered as economically profitable for heat source temperature below 100 degrees C. As illustration, a regenerator increases the system's energy efficiency by 7%, the degree of thermodynamic perfection by 3.5% while the exergetic efficiency is unchanged in comparison with the simple Rankine cycle, with R600 as working fluid. The impacts of heat source temperature and pinch point temperature difference on engine's performance are also examined. Finally, results demonstrate that energy analysis combined with the mathematical graph theory is a powerful tool in performance assessments of Rankine based power systems and permits meaningful comparison of different regenerative effects based on their contribution to systems improvements. (C) 2009 Elsevier Ltd. All rights reserved.