Models of trade policy often depend on the efficient aggregation of individual preferences. While much of the recent empirical work on trade focuses on whether domestic coalitions form along class or sectoral lines, the process of preference aggregation itself remains understudied. In democratic countries, voting is typically assumed to be an unproblematic mechanism for aggregating preferences, but such an assumption may be misleading when the salience of trade policy is low or heterogeneous throughout the electorate. Using data from a survey of 36,501 potential voters in the 2006 U.S. midterm congressional elections, this article explores the salience of trade policy for voters as a whole and for populations predicted to be most affected by changing trade patterns. The article offers an estimation of trade policy salience based on the degree to which voters held Senate incumbents accountable for their 2005 vote on the Central American Free Trade Agreement, relative to roll call votes on other issues of the day. The article finds trade policy salience to be relatively low in terms of stated importance, in voters' knowledge of their representatives' policy positions, and in its effect on voters' propensity to vote for the incumbent. The low salience of trade policy, particularly among highly affected groups, calls into question voter-driven models of trade policy.