The identification of factors affecting birth weight is a key issue in human development due to its established associations with long-term health, educational and labour-market outcomes. This paper exploits intergenerational information on three generations (grandparents-parents-children) to explore the effects of parental education and different parental smoking behaviours on birth weight. We exploit the intergenerational association between grandparents’ education and smoking behaviour and the corresponding parental variables to aid the identification of parents’ education and smoking on birth weight. We employ rich intergenerational data drawn from the US National Longitudinal Study of Adolescent to Adult Health (Add Health) and a flexible two-stage approach. We find that there is a strong intergenerational persistence of education levels and smoking behaviours across generations. Higher parental education reduces the likelihood of children’s low birth weight, although the effect appears to be mainly driven by fathers. While maternal smoking during pregnancy increases the occurrence of low birth weight, especially among non-white parents, parental regular smoking (of either mothers or fathers) does not seem to greatly affect birth weight. Results are confirmed by robustness checks excluding direct effects of grandparents’ smoking behaviour while in utero and using an instrumental variable for parental education