In this contribution, a method for the identification of a game-theoretic driver decision model is presented. The aim of the identification process is to find performance index parameters for which the solution of the optimization problem representing the driver decision making process is identical to a given set of recorded trajectories. In order to solve the identification problem a bi-level approach is used. Hereby, the lower level problem is the optimization problem modeling the decision making of the driver. The upper-level optimization problem on the other hand models the parameter optimization problem searching for the performance index parameters. A limitation of the model to be identified is, that the optimization problem modeling the decision making process cannot be solved analytically. Therefore, the identification problem can not be solved using model-based parameter-optimization techniques. To circumvent this problem an evolutionary algorithm is applied. While in the first step the algorithm is developed for the identification of a single driver, it is extended to the multi-player case in the second step. Finally, the developed approach is validated using artificially created data.