Abstract: A discrete particle swarm optimization algorithm is presented to tackle the quadratic assignment problem (QAP). Based on the characteristics of QAP and discrete variable, this paper redefines particles' position, velocity, and their operation rules. In order to restrain premature stagnation, individual-diversity of particle and average-diversity of particle swarm are defined. A repulsion operator is designed to keep the diversity of particle swarm, and an efficient local search operator is used to improve the algorithm's intensification ability. Using those operators, the proposed algorithm can get good balance between exploration and exploitation. Experiments were performed on QAP instances from QAPLIB. The simulation results show that it can produce good results.