Add to ORKGAbstract: This paper investigates the French gender gap. We estimate, on matched employees-employer data set, REPONSE 98, separated male/female earnings equations using Heckman two step procedure in order to control the potential occupational segregation (three occupational sub-samples are considered: blue collars, employees and white collars). Following Neuman & Oaxaca (2003), we decompose the estimated wage differentials taking into account for potential distributions of the selectivity. Our econometric results highlight that French women in 1998 still suffer from high pure discrimination effects (unexplained part of gender gap) whatever their occupation. In fact, all the decompositions we have adopted yield positive estimates of disc...