Abstract: In recent years, with the continuous development of railway transportation, the speed of passenger trains has continued to increase. Rolling contact fatigue failure has become the main failure form of wheel-rail materials. In order to improve the rolling contact fatigue life of train wheels, the influence of original microstructure on rolling contact fatigue life of ER9 high speed wheel steel was analyzed. It provided theoretical and experimental basis for the design and damage control of key materials of rail transit. The ER9 wheel steel with the original microstructure of lamellar pearlite + proeutectoid ferrite (P+PF) was quenched and tempered to obtain the ER9 wheel steel with the original microstructure of tempered sorbite (TS). GPM-40 rolling contact fatigue testing machine was used for rolling contact fatigue test. Because the actual wheel-rail running process is the rolling contact fatigue phenomenon after a certain dry friction operation, in order to be closer to the engineering reality, the pre-wear of the ER9 wheel steels with two different original microstructures, and then the rolling contact fatigue test is performed. In this test, ER9 wheel steel is the main sample, and U71Mn rail steel is used as the accompanying sample. Use FM-700 microhardness tester for hardness measurement, use USB digital microscope, Zeiss Supra 55 field emission scanning electron microscope with electron backscatter diffraction (EBSD) to observe and analyze the surface morphology, surface microstructure and cross-sectional microstructure of two different original microstructure samples. Analyze the reasons for the difference in fatigue life of the two samples before and after pre-wear. Results: According to the principle of colloidal equilibrium, lamellar cementite is easier to dissolve. During machining, due to the interaction of cutting force and cutting heat, The original P+PF sample will form a machined fine-grain layer, which is unevenly distributed on the surface of the sample, the maximum thickness is about 1 μm, while the original TS sample has no obvious fine-grain layer. Due to the existence of the machined fine-grain layer, the rolling contact fatigue life of the two different original microstructure samples is quite different. Rolling contact fatigue life of original P+PF sample and original TS sample are 1.6×10⁶ cycles and 5.6×10⁵ cycles, respectively. Observe the cross-sectional microstructure and cracks of the two samples after the same cycle of rolling contact fatigue test. The P+PF sample will preferentially form shallow spalling, and rolling contact fatigue cracks will be induced on the basis of shallow spalling, while the TS sample will Rolling contact fatigue cracks with an angle of 45° to the surface are directly formed. As the number of operating cycles increases, the length of fatigue cracks continues to grow. When pre-wearing two samples with different original microstructures for 1×10⁵ cycles, the surface of the P+PF sample was dominated by adhesive wear, and there was a slight fatigue wear characterized by scaly skin. While TS sample the form of wear is adhesive wear. Observing the cross-sectional microstructure of the sample at this time, both the P+PF sample and the TS sample showed a fine-grained layer due to the plastic deformation of the sample surface under the action of the contact stress. The surface hardness of the two samples has been greatly improved. The P+PF sample has a better surface hardening ability. Pre-wear causes the surface of the sample to be strengthened, which improves the rolling contact fatigue life under oil lubrication conditions.In the subsequent rolling contact fatigue test, two different original microstructure ER9 wheel steel samples all showed the shallow exfoliation of the fine-grained layer.After pre-wear, the rolling contact fatigue life of the P+PF sample and TS sample was 2.7×10⁶ cycles and 8.3×10⁶ cycles, respectively. Conclusion: During the fatigue test, the original P+PF sample will preferentially initiate shallow cracks on the surface, shallow cracks propagate in the fine-grained layer to form shallow spalling, and fatigue cracks are induced in the area where the fine-grained layer spalls. On the other hand, rolling contact fatigue cracks initiated directly after short cycles of operation of the original TS sample. Therefore, the original P+PF sample has better resistance to fatigue crack initiation. After pre-wear 1×10⁵ cycles , a small amount of fatigue wear cracks were formed on the surface of the P+PF sample, and some of the fatigue wear cracks became the crack source of rolling contact fatigue cracks. In addition, the surface hardness of the pre-wear P+PF sample is higher and more fragile, and the initiation of fatigue cracks are easy to propagate. However, due to the refinement of the surface layer of the sample, which effectively resists the initiation of cracks, the fatigue life is increased by 1.7 times compared with the original P+PF sample. After pre-wear 1×10⁵ cycles, the TS sample did not show fatigue wear cracks. At the same time, due to the surface strengthening caused by the pre-wear, the fatigue life of the TS sample was 15 times longer than that of the original TS sample, and was higher than the same cycle of pre-wear 3 times of P+PF sample.