Abstract: Mine gas monitoring system is a multi-sensor monitoring system, which measures circumstance from many azimuthes using sensors with different function, different precision and different position.But at present, there is no general and effective method to process multi-sensor data, as the result real-time and precision of mine gas monitoring system is hard to reach. So, the paper put forward a design scheme of gas monitoring system based on data fusion with BP neural network. The scheme used improved BP neural network algorithm to fuse multi-sensor data,and used two-stage fusing mode to process the data, which can gain environmental characteristics of coal mine underground. The simulation result showed that the gas monitoring system based on data fusion with BP neural network has higher measuring precision, and greatly improves reliability, integrality and efficiency of data collection.