In practical wireless system, the system state knowledge is of inaccurateness resulting from the dynamics of system states such as traffic intensity, queue status and channel condition. It is hard to make precise scheduling decision base on these inaccurate information. A wireless queue scheduling scheme based on adaptive fuzzy logic is proposed, which, by using fuzzy inference to alleviate the effect of inaccurate state information, achieves better scheduling performance. Furthermore, a reinforcement learning scheme is adopted to improve the scheduling performance. Better performance in terms of user fairness and system throughput is shown by simulation.