Abstract: In order to study the fatigue characteristics of the components under random vibration loads, the random vibration fatigue tests for the metal structure parts are carried out by using a shaking table. The stress response load of the specimen during the test are measured with the retrofitted strain gauge, and the fatigue process of the specimen is monitored in a real-time way to obtain the fatigue life accurately. Based on the measured time-domain data, and by combining the rain flow counting method with the linear cumulative damage theory, the random vibration fatigue life of the specimens under three levels is predicted. The estimated values are well within the three-fold dispersion band compared with the experimental values, which verifies the feasibility and the effectiveness of the random vibration fatigue life prediction method based on the measured time-domain data. With the method, the fatigue failure monitoring and the life prediction for structural parts in the dangerous state can be carried out through the measurement of load data.