Causal evidence of the effects of violent crime on its victims is sparse. Yet such evidence is needed to determine the social cost of crime and to evaluate the cost-effectiveness of policy interventions in the justice system. This study presents new evidence on the effects of violent crime on pregnancy and infant health outcomes, using unique linked administrative data from New York City. We merge birth records with maternal residential addresses to the locations of reported crimes, and focus on mothers who lived in a home where an assault was reported during their pregnancies. We compare these mothers to women who lived in a home with an assault that took place shortly after the birth. We find that assaults in the 3rd trimester significantly increase rates of very low birth weight (less than 1,500 grams) and very pre-term (less than 34 weeks gestation) births, possibly through a higher likelihood of induced labor. We show that our results are robust to multiple choices of control groups and to using maternal fixed effects models. We calculate that these impacts translate into a social cost per assault during pregnancy of $41,771, and a total annual cost of over $4.25 billion when scaled by the national victimization rate. As infant health is a strong predictor of life-long well-being, and women of lower socioeconomic status are more likely to be victims of domestic abuse than their more advantaged counterparts, our results suggest that in utero subjection to violent crime is an important new channel for intergenerational transmission of inequality.