Faultlines—hypothetical dividing lines splitting a group into homogeneous subgroups based on the distribution of demographic attributes—are frequently assumed to be detrimental to group outcomes because they operationalize social categorizations. However, a literature review indicates that this is not always the case. We argue that diversity faultlines and social categorizations are not necessarily the same and that the effect of diversity faultlines is moderated by perceived social categorizations. To test this proposition, we assigned 172 participants to groups of four. Participant gender, bogus personality feedback, seating position, and colored cards were employed to create two diversity faultline conditions (weak and strong faultline). Groups worked on the Survive in the Desert task, and their interactions were coded with the discussion coding system (DCS). Social categorizations were elicited using a newly developed measure that requires participants to specify subjectively perceived salient categories. Participants stated many social categories that were unrelated to surface-level characteristics frequently employed in diversity research. In line with our hypotheses, social category salience moderated the effect of faultline strength on elaboration. Elaboration was most intense in strong faultline groups that had low levels of category salience. Elaboration was positively related to performance. Theoretical and practical implications are discussed.