The ability of a multibody dynamics model to accurately predict the behavior of a real system depends heavily on the correct choice of model parameters. The identification of unknown system parameters, which cannot be directly computed or measured is usually time consuming and costly. If experimental measurement data of the real system is available, the parameters in the mathematical model can be determined by minimizing the error between the model response and the measurement data. The latter task can be solved by means of optimization. While many optimization methods are available, optimization with a genetic algorithm is a promising approach for searching optimal solutions for complex engineering problems, as reported in a paper of one of the authors. So far, however, there is no general approach how to apply genetic optimization algorithms for complex multibody system dynamics models in order to obtain unknown parameters automatically — which is however of great importance when dealing with real flexible multibody systems. In the present paper we present a methodology to determine several unknown system parameters applied to a flexible rotor system which is excited with periodic impacts. Experiments were performed on the physical system to obtain measurement data which is used to identify the impact force as well as the support stiffnesses of the rotor system using genetic optimization.