Many cities have adopted air quality alert systems to reduce the health risks from severe pollution episodes, pairing public messaging with temporary restrictions on vehicle and industrial activity. Despite their widespread implementation, evidence on their effectiveness remains mixed, in part because of data limitations and a focus on traffic-only or voluntary measures. This paper evaluates Mexico City’s air quality alert program using a fuzzy regression discontinuity design that exploits a preset ozone threshold for policy activation. I find that alerts lead to significant improvements in ozone and sulfur dioxide concentrations and sizable reductions in emergency department visits for respiratory (56 % decrease) and cardiovascular conditions (50 % decrease). The effects on transport-related pollutants are smaller and time-dependent, consistent with the alerts mitigating vehicle emissions more slowly. To assess mechanisms, I analyze information-seeking behavior, mobility data, and emissions inventories. The alerts increase online searches about air quality and the policy itself, but not about protective behaviors. Traffic volume falls and congestion improves, though public transit usage does not increase. Finally, I show that the pollution reductions are largest near restricted industrial facilities, which suggests that industrial curbs play a central role in policy effectiveness. These results can support the design of short-term environmental response policies in cities facing both mobile and stationary sources of pollution.
