We assess the lifesaving effect of hospital proximity using data on fatality rates of road-traffic accidents. While most of the literature on this topic is based on changes in distance to the nearest hospital triggered by hospital closures and use OLS estimates, our identification comes from the exogenous variation in the proximity to cities that are allowed by law to have a hospital based on their population size. Our instrumental variable results, based on Italian municipalities data from 2000 to 2012, show that an increase by a standard deviation of distance to the nearest hospital (5 km) increases the fatality rate by 13.84% on the sample average. This is equal to a 0.92 additional death per every 100 accidents. We show that OLS estimates provide a downward biased measure of the real effect of hospital proximity because they do not fully solve spatial sorting problems. Proximity matters more when the road safety is low; the emergency service is not properly organized, and the nearest hospital has lower quality standards.