Recently, as the indoor positioning technologies have been developed continuously, several indoor positioning methods have been proposed and experimented. A previous research has proposed a statistical access point selection method using a kernel density estimation (KDE) based on received signal strength indicator (RSSI). However, the proposed method is only test through a small area experiment, where the movement of the people beside the experimenters is not so much. In this paper, to analyze the stability of the proposed method under noisy environment, an experiment in a daily healthcare facility work place, where the human movement is more complex, is designed; to analyze the effectiveness of the method, several others multilateration methods are compared with the KDE method. The experiment results show the effectiveness and stability of the proposed method.