Previous research on the persuasive impact of an overheard audience has yielded conflicting results. In this study, we attempted to understand such audience effects within the framework of the heuristic model of persuasion. Subjects listened to an audiotaped persuasive message that conveyed arguments of either high or low quality and that was responded to by either an enthusiastic or an unenthusiastic overheard audience. In addition, subject involvement (high vs. low) was varied. Consistent with predictions, the audience response cue influenced postmessage opinions only under low involvement; under high involvement, only argument quality affected persuasion. Analyses that took into account subjects' need for cognition supported the additional hypothesis that individuals lower in need for cognition would be more responsive to the audience manipulation under low involvement. Thought-listing data and regression analyses provided further support for the heuristic model.