Abstract: The effects of specimen size, notch and loading type usually have important influence on the fatigue properties of metallic materials. Therefore, it is vital to modelling the effects of specimen size,notch and loading type on the fatigue strength, which is of great importance for predicting the fatigue property of the structural parts from that of the materials. In this work, rotating bending and axial loading fatigue tests are first performed on the specimens of an EA4T axle steel with different shapes. Experimental results indicate that the fatigue strength of the dogbone specimen is lower than that of the hourglass specimen due to the increase of the specimen size, and the notch reduces the fatigue strength of the specimen in terms of the nominal stress due to the large stress gradient at the notch root. The fracture surface observation by the scanning electron microscope indicates that all the tested specimens fail from the surface of the specimen. For hourglass and dogbone specimens, most of the fracture surfaces only have single crack initiation site, while for the notched specimen all the fracture surfaces have the characteristic of multiple crack initiation sites. Then, the probabilistic control volume method is used for correlating the effects of specimen size, notch and loading type on the fatigue strength of the EA4T steel, which is also compared with the critical distance method and the strain energy density method. It is indicated that the probabilistic control volume method gives better predictions for correlating the effects of specimen size,notch and loading type on the fatigue strength of the EA4T steel. Finally, a method based on the control volume is proposed for predicting the fatigue strength of structural parts, and is used for the specimens of axle steel with discontinuous highly stressed regions. The predicted results are in agreement with the experimental data.