Processes by which an attitude is polarized via social interaction were investigated in Experiment I. Subjects were led to expect interaction with the same partner or a different partner in three future sessions (high or low commitment to future interaction, respectively). All subjects received a communication from the partner which advocated either a less extreme or more extreme position than the subject's. Commitment to future interaction facilitated attitude change toward the partner's position regardless of extremity. Commitment to a consonant act (Experiment II), manipulated via public or private identification of subjects' consonant essay, also facilitated change toward an extreme consonant communication, but increased resistance to a discrepant communication. The results suggest that processes related to commitment may provide one explanation for the polarization of attitudes.In typical investigations of attitude change, one determines the reactions of the subject to a communication advocating an opinion opposite to his own. A good deal of the literature is concerned with the conditions under which a discrepant communication leads the subject to abandon his attitudinal position in favor of the opposite one.In contrast, the present research investigated the largely unexplored area of attitude change in response to an extreme but consonant communication. In this case, the extreme communication may be viewed as consonant in that it implies broad agreement with an initial attitude but specifically advocates a position more extreme than the subject's. Although relatively little is known about shifts from moderate to more extreme atti-