We contribute to the literature on relative age effects on pupils' (non-cognitive) skills formation by studying students' social network. We investigate data on European adolescents from the Health Behaviour in School Aged Children survey and use an instrumental variables approach to account for endogeneity of relative age while controlling for confounders, namely absolute age, season-of-birth, and family socio-economic status. We find robust evidence that suggests the existence of a substitution effect: the youngest students within a class e-communicate more frequently than relatively older classmates but have fewer friends and meet with them less frequently.