We study the short- and long-run effects of having a child on labour market outcomes of mothers compared to non-mothers. Using matched employer-employee data for Italy over 1985–2018, through an event study methodology around childbirth, we show that the long-run child penalty in annual earnings is 52 log points and the penalty largely depends on the reduction in weeks worked by mothers. We then investigate sorting of women with and without children across different types of firms, providing evidence that mothers work in firms with lower productivity, sales, capital and wages after childbirth. Differences in rent-sharing between mothers and non-mothers explain 11.3% of the long-run child penalty in weekly wages, mostly due to between-firm components. Finally, we explore the individual-level, firm-level, and cultural factors that influence the size of child penalties. We find that the child penalty is higher for young, low-wage mothers and those taking longer leaves. It is larger in small firms with less generous pay and worse peers, and in more gender-conservative regions.