A two-dimensional BEM model of a surface initiated rolling contact fatigue crack has been developed. The model includes the effects of Coulomb friction between the faces of the crack, and the possibility of fluid pressure acting on the crack. The model has been used to investigate three different mechanisms to propagate contact fatigue cracks. In the fluid lubrication mechanism, it is assumed that no fluid pressure acts on the faces of the crack or the fluid in the crack merely lubricates the crack faces, and reduces the crack face friction coefficient. Under these conditions, the crack virtually did not open, so that only a negligible KI was generated at the crack tip. This mechanism therefore relies on the assumption that contact fatigue cracks are propagating in mode II driven by cyclic shear stresses. The BEM model has been used to calculate the cycle of KII at the tip of a pitting crack as the load rolls over it. However, as the attempts failed to grow mode II cracks in structural steel experimentally, this does not seem to be a mechanism to propagate contact fatigue cracks in rail and wheel steel. The hydraulic pressure mechanism has also been investigated. In this mechanism, fluid is forced into the crack by the load, so that the faces of the crack are prized apart. In the case of this model, when the Hertzian contact pressure just sealed the mouth of crack and pressure that was equal to the magnitude of the edge of the Hertzian pressure was applied to the crack faces, the mouth of the crack shut for the certain condition because the edge of the contact pressure was very small. When the mouth of the crack shut, it was impossible to consider the fluid transmitted the contact pressure to the crack faces and the calculation could not proceed. Therefore, a way to deal with the case when the mouth has shut should be considered to establish a consistent model. Finally, the fluid entrapment mechanism has been considered. The resulting cycles of KI and KII, have been calculated as the load rolls over the crack. lt should be noted that experiments by using similar loading cycles have successfully grow long co-planar cracks in structural steels. The effect of changing parameters on the cycles of stress intensities is as follows: (i) The effect of changing crack face friction is not so significant and the variations of KI and KII are almost the same. (ii) The changes of surface traction have a relatively strong influence on KI. lf the driving traction increases the crack mouth is pulled open more to trap more fluid in the crack. As a result, the value of KI increases. (iii)The changes of crack length have a strong influence both on KI and KII. If the crack length increases, both of KI and KII increase with their profiles kept almost the same. (iv)The effect of changing crack inclined angle is significant. When the angle decreases, the value of KI increases and the variation patterns of KI are significantly different from one another. The easiness of crack growth and its direction according to the crack inclined angle were also discussed in detail by comparing some recent experiments and the numerical results in this study.