because of the poor ability of different environmental adaptation,the measurement distance exist errors in traditional RSSI (received signal strength) positioning method,which directly affect the positioning accuracy of the wireless sensor network. for the defect, a RSSI-based neural network positioning method is proposed . Building the mapping relations of the unknown node RSSI value and coordinate, then locating with neural networks. the method reduces influence of different environments for RSSI,which makes the algorithm have a certain robustness. The experimental results show that compared with the traditional RSSI localization method,this algorithm has definitely improved. in positioning accuracy.