Indoor positioning plays an important role in many applications, such as public safety, healthcare, location-based services and so on. How to improve positioning accuracy and the model' adaptivity to the environment becomes a key issue of indoor positioning, where most existing techniques generally use the value of the received signal strength indication (RSSI) to obtain the distance. Given the fact that the traditional logarithmic distance path loss model cannot adapt to the complex indoor environment very well, we propose a context-adaptive segmentation heterogeneous RSSI fitting positioning method. The proposed method firstly utilizes the difference in signal transmission under different application scenarios to divide the RSSI data into several different segments. Then it finds the optimal piece-wise fitting point by RSSI's differentiated characteristics, and selects the optimum function for each seg-ment, enabling the number of segments, segment position, and piecewise functions of all segments to a-dapt to corresponding application scenarios. Finally high accurate RSSI signal fitting is achieved. Exper-imental results show that the proposed method in RSSI fitting can achieve higher accuracy than the traditional single-fitting function, and improve the accuracy of position algorithms significantly.