Abstract: I link 6-hour air pollution exposure to the total number of car accidents in the city of Santiago by exploiting time-series variation from 2013 to 2016. In order to identify the causal effect of CO exposure, I use plausible exogenous variation in atmospheric stability to instrument CO exposure. I found a nonlinear relationship between CO exposure and the total number of car accidents. This result is driven by nonfatal accidents. Indeed, I do not find any impact on fatal accidents. In addition, the results hold under a battery of robustness checks. Although Santiago’s CO level is far below the international criteria of a hazardous level, I argue that reducing the average level of pollution leads to a sizable increase in social welfare due to a reduction in the number of car accidents.