Optimizing the load spectrum extraction method is the key to improve the accuracy of fatigue analysis of automotive suspension components.Taking a compact SUV and a Middle Class as research objects, a rigid-flexible coupled multi-body dynamic model of the whole vehicle was established by using the ADAMS software with the parameters obtained by field tests.The acceleration and displacement response signals at the multi-position of suspension and the six-component force at wheel center were obtained by tests and measurements on a proving ground as well as on a real vehicle test rig.The virtual iteration method and body constrained method were used to simulate the working conditions of the test rig, and the response data at the monitoring positions of the multi-body model were obtained.The simulation and test curves at suspension monitoring positions were analyzed and compared in terms of the time domain signal and root mean square value.The results show that the body constrained method is more suitable for high frequency and low amplitude vibration conditions, and as contrast, the virtual iteration method has wider application scope and higher simulation accuracy, but is more time-consuming.The above conclusions were verified by the working tests on the washboard road and the long wave road in the proving ground.At the same time, the load spectrum at the steering knuckle was extracted by the above two methods, and the pseudo-damage value of the load spectrum was calculated.It is found that the fatigue life of the steering knuckle estimated by using the load spectrum extracted by the body constrained method is more conservative than that by the virtual iteration method.