This paper uses a large multi-country database with data from the OECD PISA program to disentangle the effects of birthdate on educational performance. As far as age effects are concerned, we conclude that children are disadvantaged because they are the youngest in class (relative age effect), not because they are young per se. Our findings go against delaying mandatory school entry as a general policy, as there is no gain from a rise in entry age - keeping age differences among students constant - to make up for the shortening of length of schooling. This evidence that postponing school entry postpones learning is more marked for children belonging to disadvantaged households. In contrast, the relative age effect does not interact with family background, and remains stable across school entry age cohorts. The size of this effect, measured at the age 15 is not large, but its interaction with early grade retention and tracking may enhance long-term effects. Finally, we do not detect an association between birthdate and achievement originating in unobservable characteristics of students.