To monitor mine gas better, extract the abnormal gas data rapidly and eliminate the normal data redundancy, this paper proposes a WSN data fusion based on cluster method. It used the way of minimizing the total energy consumption that all nodes are devided into clusters best, nodes in each cluster determined the cluster head through the election of cluster head probability, the cluster nodes judged the data collected whether to be abnomal or not through the two thresholds and sent it to the monitoring stations for staffs to decide. The simulation results show that this method prolongs the wireless sensor network lifetime, the mine gas is monitored effectively.