Studies of political participation and representation often contend that elected officials respond more to the preferences of voters than those of nonvoters, but seldom test this claim. This is a critical assumption because if true, biases in who participates will lead to biased representation. Office holders might respond disproportionately to voters’ preferences because voters tend to select like-minded representatives, voters tend to communicate their preferences more, and only voters can reelect representatives. We find that voter preferences predict the aggregate roll-call behavior of Senators while nonvoter preferences do not. We also present evidence supporting the three explanations advanced to account for the preferential treatment of voters.