We study whether referrals deliver better information about the productivity of new hires than other recruiting channels. Using combined vacancy-register data, we document that referrals are associated with less screening and lower search costs. We demonstrate that the referral wage premium is informative on referrals’ information advantage only if differences in screening are accounted for. Consistent with the predictions from a learning model, referrals increase new hires’ job stability and wages, with the effects diminishing over time. Results suggest that the referral information advantage is particularly pronounced for males and when employers use referrals as the only search channel. • Referrals are associated with less screening and lower screening costs. • Referral wage premium reflects information advantage after conditioning on screening costs. • Referrals increase new hires’ entry wages and job stability. • Effects diminish over time. • Empirical analysis supports predictions of a learning model.