Parameter identification of chaotic systems is a key problem in chaos control and synchronization of nonlinear science. An improved quantum genetic algorithm with good global search capability is proposed, and it is used for parameter identification of chaotic systems. The fitness function is constructed by reducing the state synchronization error between the actual system and mathematical model. The parameter identification problem is transformed into a multi-dimensional optimization problem. The hyper-chaotic Chen system is investigated with the improved algorithm, and it is compared with the basic quantum genetic algorithm. Results show that the improved algorithm has better optimizing ability, and it is proved that the improved quantum genetic algorithm is effective and feasible.