In the present study, a multi-objective optimization procedure combining finite element modeling of impingement cooling system, Response Surface Approximation (RSA) of objective functions and optimization based on Multi-Objective Genetic Algorithm (MOGA), to achieve maximum heat transfer and minimum entropy generation is demonstrated. For the purpose, numerical simulations are performed for impingement cooling system with Al2O3/water nanofluid, to investigate the influence of Reynolds number (Re), non-dimensional channel height (H/L) and nanoparticle volume fraction (Pdbl) on fluid flow, heat transfer and entropy generation. The simulated results illustrate that, a secondary recirculation bubble observed on upper surfaces of fourth (at Re ≥ 500) and fifth (for Re ≥ 800) heat sources, lead to an accumulation of heat. The magnitude of local Nusselt number (Nu) is found to be maximum along stagnation region whereas in the regions of secondary recirculation a minimum value is observed. Further, an increase in overall surface averaged Nusselt number (Nuov) and global total entropy generation (Stot,Ω) is observed with increasing Re,Pdbl and decreasing H/L. Subsequently, Nuov and Stot,Ω are selected as objective functions and are modeled using RSA. Furthermore, MOGA has been implemented to obtain optimum configurations of impingement cooling system encapsulating in the functional space lying on the Pareto-optimal frontier where a trade-off between two performance parameters, Nuov and Stot,Ω are obtained. © 2015 Elsevier Ltd. All rights reserved.