Intergroup interactions are often anxiety provoking, and this can lead members of both majority and minority groups to avoid contact. Whereas negative consequences of experiencing intergroup anxiety are well documented, the role of perceived anxiety has received substantially less theoretical and empirical attention. We demonstrate in 3 experiments that the perception of anxiety in others can undermine intergroup interactions even when the anxiety can be attributed to a source that is unrelated to the interaction. Participants who learned that a cross-race partner's anxiety could be attributed to an upcoming evaluation (Study 1) or a stimulant (i.e., caffeine, Studies 2 and 3) expressed less interest in continuing an interaction (Studies 1 and 2), showed less self-disclosure (Study 2), and increased physical distance between themselves and their partner (Study 3) than did those given no source information and participants who interacted with a same-race partner. Moreover, compared to control participants, perceivers who were given an incidental explanation for their partner's anxiety perceived outgroup, but not ingroup, partners as more anxious (Studies 1 and 3) and showed heightened accessibility of anxiety words (Study 3), indicating that incidental source information enhanced accessibility of intergroup (but not intragroup) anxiety at early stages of information processing. Theoretical and practical implications for combating paradoxical effects of perceived anxiety in intergroup interactions are considered.