An emerging literature finds that early life exposure to conflict has important effects on subsequent physical and cognitive development. While this literature focuses on large-scale violent events and low intensity conflicts, there is a lack of studies examining high levels of criminal violence. This discrepancy is important as many areas in the world, particularly Central and South America, experience consistently high levels of organized crimes. This study examines whether these health effects also extend to criminal violence setting by focusing on the sharp increase in homicide rates in Mexico since 2007-08. Using sibling fixed effects, I study whether the levels and timing of municipality homicide rates affect children’s physical health and cognitive and non-cognitive development in Mexico. The results show a strong effect of in utero exposure (depending on the trimester) on the physical health and cognitive development and no effect on socio-emotional behavior and chronic illnesses. Specifically, an average increase in the homicide rate between the pre-escalation period of 2005-06 and 2009 while in utero reduces both height- for-age Z-scores (HAZ) and cognition (measured by Raven’s scores) by 0.08 standard deviation (SD). The results further provide suggestive evidence about maternal stress and prenatal care use as potential channels.