The present paper provides new evidence that hospital delivery can significantly lower child mortality risks, especially among vulnerable young adolescent mothers in Bangladesh. We exploit the exogenous variation in community's access to local health facilities (both traditional and modern) before and after the completion of the 'Women's Health Project' in 2005 (that enhanced emergency obstetric care in women friendly environment) to identify the causal effect of hospital delivery on various mortality rates among children born during 2002-2007. Our best estimates come from the parents fixed effects models that help limiting any parents-level omitted variable estimation bias. Ceteris paribus, access to family welfare clinic boosts hospital delivery likelihood, which in turn tends to lower neo-natal, early and infant mortality rates, especially among adolescent mothers after the completion of Women's Health Project.