We study the persuasive effects of political advertising. Our empirical strategy exploits FCC regulations that result in plausibly exogenous variation in the number of impressions across the borders of neighboring counties. Applying this approach to detailed data on television advertisement broadcasts and viewership patterns during the 2004–12 presidential campaigns, our results indicate that total political advertising has almost no impact on aggregate turnout. By contrast, we find a positive and economically meaningful effect of advertising on candidates’ vote shares. Taken at face value, our estimates imply that a standard deviation increase in the partisan difference in advertising raises the partisan difference in vote shares by about 0.5 percentage points. Evidence from a regression discontinuity design suggests that advertising affects election results by altering the partisan composition of the electorate.