To solve the shortcomings of sampling efficiency and positioning accuracy of the Monte Carlo localization algorithm in Wireless Sensor Networks (WSN), a Monte Carlo localization Boxed (MCB) algorithm for mobile nodes based on Received Signal Strength Indication (RSSI) ranging was proposed. To improve the positioning accuracy, the filter conditions was strengthened by mapping the ranging information into different distance intervals. At the same time, the samples which had already met the filter conditions were used to create more effective samples so as to improve the sampling efficiency. Finally, the Newton interpolation was used to predict the nodes' trajectory. The closer the trajectory between the sample and the node is, the greater the weight of the sample is, and the best estimate position could be obtained with these weighted samples. The simulation results indicate that the proposed algorithm has good performance in different density of anchor node, communication radius, and movement velocity etc., and compared with the MCB algorithm, the proposed algorithm has higher positioning accuracy.