Rolling contact fatigue in rolling element bearing (REB) is a common surface degradation and failure mechanism. Limited experimental studies were reported on fatigue degradation, in particular, in grease lubricated conditions. This project examines the evolution of the fatigue degradation process through collecting, imaging and characterisation of the damaged surface to further understand the tribological changes in the process. Ball bearings with a seeded, irregular-shaped dent on the outer race were tested in a grease lubricated condition and on a bearing test rig. Moulding technique was used to replicate the damaged surface during the fatigue test. Also, 4 different shapes of initial defects and round-shaped dents in 3 sizes were introduced onto the outer raceway of test bearings to study effects of the shapes and sizes of initial defects on the fatigue degradation process as they are considered significant factors in fatigue propagation. This study has confirmed that the fatigue damage propagated on both the trailing and leading edges of the initial defect. The quantitative analysis reported that the wear depth on the trailing edge was measured more than that on the leading edge by over 100 μm, indicating different stresses distribution on the two edges of the seeded defect. In addition, the leading-edge damage surface was observed ‘self-healing’ or smoothening phenomenon. The shape study suggested similarities with propagating behaviour among the tested shapes of the initial defects. The initial shapes of defects mainly affected the fatigue propagation at the early stage where the first spall was initiated and fatigue progression on the leading-edge side of damaged surface in terms of the length of defected surface and the features of the damaged surface. The size influence study demonstrated that increasing size of the defect enlarged the effect of impact by rolling elements, evidenced by generating rougher surface, and thus shortened fatigue life. The study provides insights in the fatigue process of rolling bearings through monitoring the evolution of the tribological features of the degraded surface. The obtained information can be useful in fatigue model validation and development of effective online techniques for monitoring and prediction of the fatigue progress of REBs.