Criteria used in hiring workers often do not reflect skills required on the job. By comparing trainee performance for newly hired workers conditional on competitive civil service examination scores for hiring French public sector workers, we test whether women and men with the same civil service examination score also exhibit similar performance in the job-related trainee program. Both civil service examinations and trainee scores contain anonymous and non-anonymous components that we observe separately. We find that by the end of the trainee program (first year of employment), women are outperforming men on both anonymous written and non-anonymous oral evaluations, a finding that holds both conditionally and unconditionally on civil service examination results. We also note a minimal correlation between the competitive exam content and trainee performance, which implies that the former give women insufficient opportunity to reveal their job-related skills. This design flaw results in functional discrimination against women, although the hiring process seems procedurally fair in that women’s and men’s non-anonymous oral civil service examination scores hardly differ after controlling for the performance in the anonymous written scores.