In contrast to early studies of voting behavior, where selective exposure was proposed as an explanation for limited media effects, this dissertation contends that selective exposure is a cause of potentially significant media effects. This study documents the extent of exposure to politically congenial outlets and identifies some of its key causes and consequences. Data from the 2004 National Annenberg Election Survey are used to examine the contours of partisan media use, supplemented by an experiment investigating whether the media environment's structure influences partisan selective exposure. The results offer strong evidence that people choose political media in accordance with their political predispositions, and that political interest and knowledge are prerequisites for selective exposure. Media offerings appear to matter: findings suggest that when people have more media options from which to choose, their long-term exposure decisions are more apt to be biased toward congenial media. Overtime survey analyses suggest that salient political media events may encourage selective exposure. Turning to the consequences, analyses provide support for the view that partisan selective exposure contributes to political participation, limited evidence that it leads people to settle on their vote choice earlier in the campaign, and strong evidence that it leads to higher levels of political polarization. Partisan media use also appears to contribute to differentiated patterns of agenda setting, such that audience members adopt different issue priorities depending on their news exposure. Limited evidence supports the idea that partisan media use primes the use of different issues in judging the president's performance. Results are discussed in light of two contrasting views of partisan media use in writings on communication and democracy. On one hand, partisan selective exposure inspires citizen participation and facilitates a partisan schema for making sense of the political world. On the other hand, it polarizes opinions and fragments the public. This dissertation proposes that, to the extent that the partisan media use is counterbalanced by forces that unite people into a public, it can serve a democratically beneficial role. The explosion of partisan outlets today and the decline of news outlets garnering diverse national audiences, however, warrant critical attention.