How do events that highlight a policy issue impact political preferences? In this paper, I analyze the impact of mass shootings on voter behavior. I show that, conditional on population, mass shootings are largely random events. Using a difference-in-differences strategy, I find that mass shootings result in a 1.7 percentage point loss in Republican vote share in counties where they occur. Identification that relies on comparing successful and failed mass shootings yields similar results. Mass shootings lead to an increase in the salience of gun policy and increase the divide on gun policy among both voters and politicians. Democrats (Republicans) tend to demand even stricter (looser) gun control after mass shootings. These results suggest that increasing the salience of an issue may polarize the electorate.