This paper studies the effects of talk radio, specifically the Rush Limbaugh Show, on electoral outcomes and attitude polarization in the U.S. We propose a novel identification strategy that considers the radio space in each county as a market where multiple stations are competing for listeners' attention. Our measure of competition is a spatial Herfindahl-Hirschman Index (HHI) in radio frequencies. To address endogeneity concerns, we exploit the variation in competition based on accidental frequency overlaps in a county, conditional on the overall level of radio frequency competition. We find that counties with higher exposure to the Rush Limbaugh Show have a systematically higher vote share for Donald Trump in the 2016 and 2020 U.S. presidential elections. Combining our county-level Rush Limbaugh Show exposure measure with individual survey data reveals that self-identifying Republicans in counties with higher exposure to the Show express more conservative political views, while self-identifying Democrats in these same counties express more moderate political views. Taken together, these findings provide some of the first insights on the effects of contemporary talk radio on political outcomes, both at the aggregate and individual level.