Abstract. Towards the problem of the time-varying AP signal strength degrades the indoor positioning accuracy in Wireless Local Area Network (WLAN), a WLAN indoor positioning method based on multiple mixed distribution model (MMDM) and Adaboost algorithm is adopted. Firstly, in order to describe the probability density distribution of AP signal strength accurately and improve the system suitability, the proposed method employs Gaussian mixture model, Binomial mixture model and Poisson mixture model to make up the MMDM and construct the fingerprint database. Secondly, in order to avoid the insufficient training of positioning model caused by lacking of training samples, the Adaboost is used to comprise the weak classifier based multiple MMDM into a strong classifier and improve the generalization ability of the system. Lastly, the mapping relation between fingerprint data and real position is also built by Adaboost classifier in online positioning stage. The experimental results show that the proposed method is superior to several indoor positioning algorithms with better time shifting adaptability and positioning accuracy.