Abstract: In 802.11 wireless networks, the nodes which belong to different networks are rational and selfish, which results in low fairness, low load-balance and low social efficiency of variable-width channel allocation mechanisms. In this paper, we study the variable-width channel allocation problem from a non-cooperative game-theoretic point of view in 802.11 wireless networks. Firstly, we model the variable-width channel allocation problem as a strategic game, and prove the existence of the Nash equilibrium (NE) strategy, and show the conditions that guarantee the variable-width channel allocation process converges to the NE state. Secondly, we propose an incentive mechanism based on payment to cope with the social inefficient problem of the NE strategy. The incentive mechanism influences the nodes' allocation behavior and enables the system to converge to the dominant strategy equilibrium (DSE) state, in which the performance of the whole system attains the global optimality in terms of system-wide aggregate throughput. Meanwhile, we consider and analyze the fairness and load-balance of both two strategies. Finally, we propose two variable-width channel allocation algorithms to achieve the NE and the DSE states. We have evaluated the efficiency of two algorithms and discussed the influence of the proposed schemes on the system-wide throughput. Simulation results show that the NE strategy achieves good fariness, and the DSE strategy works better than the NE strategy in terms of load-balance and social efficiency.