The result of design was affected by the variation of parameters, and the model of robust optimal design was established with the robust optimal design theory. Considering the variation in the mean value and variance of objective function and constraint function, the optimization model was built, which took decoupling of energy distribution as an objective, and stiffness parameters of mountings as a design variation. Particle swarm optimization (PSO) was adopted to optimize an engine mounting system to the stiffness of mountings, and the Monte Carlo method was used to analyze the optimization values. An example showed the method presented could decrease the sensitivity of decoupling with respect to the stiffness of mountings, and could make the vibration coupling distribution more reasonable. Comparing with the genetic algorithms, the optimization design computational efficiency of engine mounting system could be obtained quickly by PSO.