Localization of nodes is an important application of wireless sensor network. In order to suppress effectively the effects of RSSI error to WSN node localization precision. In this paper, RSSI-based distance estimated model was studied and analyzed, and provide a parameter calibration method based on the least square algorithm. The method uses least square method to amend the data to dispel the influence caused by all kinds of molestation. Thus the positioning accuracy of entire network can be improved. Simulation results show that the proposed calibration method performs well in distance estimate accuracy.