Pre-print, Working paper: This paper combines detailed household-level data on child health with geo-coded incidences of vi- olence in Iraq to estimate the impact of in-utero exposure to violence on biometric, behavioral and cognitive outcomes of children. Rich data on severity (duration and casualties), type (bombings, ex- plosions, gun_re etc.), and perpetrators of violence (coalition, insurgent, or sectarian) on the district level allow me to discriminate between two possible mechanisms: damages to the infrastructure versus violence-induced pre-natal stress for mothers. Comparing siblings within the same household, I find that one single violent incidence during pregnancy significantly increases the risk of stuntedness, malnutrition and weakens major cognitive and behavioral skills. While the type of violence does not seem to play a major role, the perpetrator of violence seems to matter. Violent acts that explicitly target the civilian population, even if they have little e_ect on the general infrastructure, appear to be the driver behind the effect.