Between April and July 1994 Rwanda experienced a tremendous wave of inter-ethnic violence that caused at least 500,000 deaths. Combining birth history data drawn from the 2000 Rwanda Demographic and Health Survey with prefecture-level information on the intensity of the conflict, we examine the impact of the civil war on infant and child mortality. War exposure is measured exploiting the differential effects of timing of birth and genocide intensity at the household and geographic level. Considering both in utero and postnatal war exposure, we estimate discrete time proportional hazard models of child mortality for the exposed and the unexposed birth cohorts. We find large positive effects of exposure to the conflict on infant and child mortality. Moreover, restricting our sample to the survivors, we find that child mortality is significantly impacted by war exposure, increasing the hazard rate by nearly 6 percentage points on average. This result holds true also for children who were only exposed while in utero. This evidence points to the existence of long-term disruptive effects on the cohorts of children exposed to the violence.