The events of 9/11/2001 have stimulated an unprecedented degree of attention to, and efforts to identify, individuals of Arab descent. Following the Holocaust, social psychologists studied the extent to which Jewish and Non-Jewish people could accurately categorize faces as Jewish vs. Non-Jewish. This study provides a systematic analysis of the results of these earlier studies of the categorization of Jewish faces, with an eye toward informing the current cultural fixation upon the categorization of Arab faces. A meta-analysis was conducted upon all available tests of the accuracy of categorization of Jewish and Non-Jewish faces, allowing the empirical scrutiny of three models of facial perception. Judges can categorize Jewish faces to a significant, albeit extremely small degree; judges can categorize Non-Jewish faces to a much greater degree. Results are most consistent with a model of facial perception that incorporates variations of cognitive representations of groups as a function of group size. These results suggest that training protocols for airport security guards should engage exemplar cognitive representations, emphasizing distinguishing features (rather than ethnic category membership) of faces. Copyright © 2003 John Wiley & Sons, Ltd.