To quantitatively estimate rolling contact fatigue strength as a crack problem, rolling contact fatigue tests were performed on JIS-SUJ2 bearing steel using specimen plates into which small holes of various diameters and depths had been drilled. In all the tests, fatigue cracks initiated at the edge near the bottom of the hole, and then propagated by the shear-mode. Even in the specimens that did not fail after being tested up to N = 2×10 8 cycles, short fatigue cracks were found at the edge. The fatigue life N f plotted against the maximum contact pressure q max varied greatly on the basis of the diameter and depth of the hole. The effect of the depth of crack initiation on Nf was uniquely characterized by using the nominal shear stress amplitude τ a , instead of q max Further, by considering the rolling contact fatigue strength as a small shear-mode crack problem, the fatigue life data was plotted using the novel parameter, τ a /(√area) −1/6 , where the area is a projected area of the hole. Consequently, all the fatigue life data was successfully fitted to a unified line irrespective of the diameter and depth of the hole, i.e., a defect size dependence on the rolling contact fatigue strength was manifested in a small crack regime.