The purpose of this investigation is to clarify the role of roughness on rolling contact fatigue. Tests have been carried out on a two-disk machine, for two rolling bearing steels (52100 and M50), two surface roughnesses corresponding to EHL and micro-EHL conditions (two different surface finishing), three normal loadings (1.5, 2.5 and 3.5 GPa), and under pure rolling or rolling plus sliding conditions. No surface damage has been observed up to 50 10(6) cycles for tests with smooth specimens. Tests with rough specimens have produced a typical surface damage, called here surface distress, made of a fat ge population of asperity-scale micro-cracks and micro-spalls. The paper describes the surface distress observed such as micro-cracks and micro-spalls, Surface damages obtained are different for tests under pure rolling conditions and rests under rolling plus sliding conditions. Therefore, the role of the friction direction is underlined. A link is made between our experimental observations and calculations that have been carried our using a transient EHL model. The influence of an indent in a line contact, simulating a micro-spall, is studied. Surface pressure and associated sub-surface stress field are analyzed versus the sliding direction.