The scientific and perfect early warning system for tunnel structure health monitoring has not been given by domestic and overseas; so the efficacy coefficient method is introduced to complete weighted comprehensive of multiple sensors monitoring data obtained from the health monitoring system to realize real-time warning or evaluation of the tunnel structure stability. Combining with the characteristics of health monitoring system, the method using support degree of the sensor monitoring data at the same time to determine the early warning index system (the optimal fusion data set) is proposed to solve the difficult problem of selecting index system in efficacy coefficient method. The three-scale method is chosen to determine the index weight which effectively overcomes the disadvantage of the 1-9 scaling method which has strongly subjectivity and the poor consistency judgment matrix. Applying the proposed method to the warning system of Wuhan Yangtze River tunnel health monitoring, firstly the monitoring data of which the support degree are greater than 0.1 is selected from the four surface strain gages and the six surface joint gages to constitute the early-warning index system; then the three-scale method is used to determine the index weight; finally the total efficacy coefficient value is calculated to complete tunnel early-warning; and the early warning results are in conformity with tunnel actual operating conditions; so the efficacy coefficient method has a good applicability in early warning of tunnel structural health monitoring.