TC4 titanium alloy is usually used to manufacture engine blades, blings or blisks and fatigue is the main failure of these components due to its high strength, good corrosion resistance and light weight. In engineering applications, three typical surface modification processes such as shot peening (SP), laser shock peening (LSP) and low plasticity burnishing (LPB) were employed to improve fatigue performance. In this work, SP, LSP and LSB were taken to enhance surface layer of TC4 titanium alloy. The surface integrity of specimens including surface roughness, microhardness, residual stresses and microstructure was investigated to obtain the effects of modification on surface layer by different methods. The rotating-bending fatigue performance was tested at room temperature and fatigue fracture surfaces were analyzed by SEM. Fatigue life was compared at the same stress 760 MPa with the reference machinced specimen. Fatigue strength was determined by stair method for 1×10⁷ cyc. The results show that both the rotating-bending fatigue life and fatigue strength of TC4 titanium alloy are increased by these surface modification processes. The fatigue life prolonging factor (FLPF) for SPed specimens is 20.4, and FLPF for LSPed specimens and LPBed specimens is 89.6 and 99, respectively. Meanwhile, fatigue strength improvement percentage (FSIP) for SPed, LSPed and LPBed specimens is 36.3%, 37.8% and 38.8%, respectively. Moreover, the fatigue cracks initiate beneath surface enhanced layer for surface-modified specimens, while they are located at surfaces for un-surface-enhanced ones. Based on dislocation theory, the subsurface cracks initiation resistance and fatigue strength for surface-enhanced specimens were analysied. Finally, surface modification mechanisms were discussed and some quantitative analysis methods on surface modification effects were proposed. For surface-enhanced smooth specimens, the FSIP limit is 40% based on proposed analysis model and it is verified in this work by different surface layer enhancement processes (36.3% for SPed specimens, 37.8% for LSPed and 38.8% for LPBed specimens are near to 40%). Fatigue total life including initiation and propagation is a complex problem, and therefore it is difficult to give accurate life prediction and analysis, especially for small crack growth, although some invesitigations on total fatigue life can be roughly estimated based on Basquin relation for stress fatigue life or Coffin and Marson eqution for strain fatigue life which have not any physical meaning or any mechanism.