The hypothesis that a political candidate's image is not constant and differs predictably among different subgroups of the electorate was tested. Three different samples of undergraduate subjects were used in a 2 × 2 factorial design in which equal numbers of Republicans and Democrats rated the images of Gerald Ford and Jimmy Carter by means of the semantic differential just prior to the 1976 Presidential election. As hypothesized, in all three samples, Republicans rated Ford higher than Carter, while Democrats did just the reverse. Though Democrats rated Carter higher than Ford, more of them indicated an intent to vote for Ford than for Carter. This finding contradicts previous information that image ratings predict voting behavior but may represent a special case in that college students' weak party affiliation, their lack of knowledge about politics, and their responsiveness to their agemates may make them extremely susceptible to peer pressure in elections.