We estimate the impact of exposure to conflict on health outcomes using geographic information on households' distance from conflict sites – a more accurate measure of shock exposure – and compare the impact on children exposed in utero versus after birth. The identification strategy relies on exogenous variation in the conflict's geographic extent and timing. Conflict-exposed children have lower height-for-age, and impacts using GPS information are 2-3 times larger than if exposure is measured at the imprecise regional level. Results are robust to addressing endogenous migration. Health service disruptions and maternal stressors are potential explanations for the negative health effects on children.