In this paper, we propose an efficient algorithm for localization of underground miners using RSSI and ZigBee. RSSI is widely used to measure the distance in wireless system. However, there are possible interferences that may affect the result of measurement, such as propagation, reflection, etc. In this paper, we develop an algorithm that can increase the accuracy of localization especially in underground applications. Gaussian distribution is used to remove the event with high effect but low possibility. After that, the statistical average is calculated. Then, three checking points are used to measure the parameters in a tunnel. Finally, the parameters are updated to the system in real time. The results show considerable improvement compared with other methods.