In this paper multi-objective genetic algorithms were employed for Pareto approach optimization of turboprop engines. The considered objective functions are used to maximize the specific thrust, propulsive efficiency, thermal efficiency, propeller efficiency and minimize the thrust specific fuel consumption. These objectives are usually conflicting with each other. The design variables consist of thermodynamic parameters (compressor pressure ratio, turbine temperature ratio, Mach number) and propeller geometric parameters (blade activity factor and integrated design lift coefficient). Main effect of the design variables are calculated to recognize which design variables has the effect on the objective functions. Group method of data handling (GMDH) type neural networks is used for modeling and prediction of propeller efficiency using aerodynamic variables obtained by some experimental data. Relationships among design variables and optimal objectives functions have been obtained by non-dominated sorting genetic algorithm, (NSGA-II) with a new diversity preserving mechanism in multi-objective optimization. The pareto solutions are obtained for both two and five objective optimization processes. For two-objective optimization, different pairs of objectives have been selected. More ever, these objectives have also considered for a five-objective optimization problem. Variables based on this pareto front, indicated the best design point of objective functions. These results also showed that Pareto solutions of five-objective optimization provide more choices for optimal design of turboprop engines.