There is a stark socioeconomic gradient in breastfeeding, which is far more common amongst the relatively more educated. However, whether breastfeeding improves child development remains an open question. We use variation in timing of birth within the week to estimate the causal effects of breastfeeding on various dimensions of children’s development. We argue that timing of birth affects breastfeeding due to variation in the provision of infant feeding support at hospital throughout the week. We provide evidence that, for normal deliveries at least, timing of birth is random. We also show that a large range of observed characteristics, including those relating to labor and delivery, are extremely similar by timing of birth. We estimate the effects of breastfeeding on overall indicators of children’s cognitive and non-cognitive development and health up to the age of 7, for a sample of low educated mothers. We find that breastfeeding has large effects on children’s cognitive development. On the other hand, we find no evidence that breastfeeding affects non-cognitive development, or children’s subsequent health. These findings are in line with the results from the only randomized trial on the topic. The paper also sheds light on the mechanisms through which breastfeeding affects development, considering how it interacts with other investments in children, maternal mental health and the mother-child relationship. Alongside several robustness tests, we perform a Monte Carlo exercise which shows that our instrumental variable estimates are conservative and the standard errors are of the right size. (Coauthor Emla Fitzsimons.)