To achieve a high performance, the design optimization of lumped parameter for Magneto-Rheological (MR) engine mount is essential. Mathematical model of single degree vibration isolation system was established and multiple interval sensitivity method is proposed to overcome drawbacks of conventional optimization design such as single objective optimization, improper optimization objective, unfeasible machining and so on. Optimization variables in MR engine mount are selected by lumped parameter multiple interval sensitivity analysis. The integral of force transmissibility through normal frequency range of engine is assigned as objective functions, non-dominated sorting genetic algorithm (NSGA-II) is improved and used to optimize design variables. The synthesized distance between Pareto lumped parameters and discontinuous lumped parameters that match along with the physical discretization dimension were calculated to select the most appropriate solution from Pareto lumped parameters.