Abstract: Node self-localization is one of the important research topics in WSN. APIT is a major range-free localization algorithm. Compared with other range-free algorithms, APIT can achieve higher precision position estimation with small communication cost. However, APIT requires high anchor node density. Besides, in the process of APIT test, boundary effect and low neighbor node density can easily increase InToOut error and OutToIn error. Otherwise, the grid scan algorithm is inefficient and has a lower fault-tolerance to OutToIn error. In allusion to the problems mentioned above, an improved APIT algorithm based on triangle-center scan is proposed. Firstly, the reason for InToOut error and OutToIn error is analyzed and two improvements of APIT are introduced. Then, the effect of grid scan algorithm on the precision of position estimation and the algorithms efficiency are analyzed, and a triangle-center scan algorithm is presented. Finally, simulation results show that the improved algorithm not only can reduce the IntoOut error and OutToIn error effectively and improve the precision of position estimation, but also has a higher fault-tolerance to OutToIn error and can enhance the algorithms efficiency.