Using a longitudinal approach, we examined intergroup bias based on randomly assigned novel groups in fourth- and fifth-grade children (roughly 9–11 years old) involved in an afterschool program. We investigated not only the development of intergroup bias but also its persistence over multiple weeks. Our intergroup bias measure assessed children's evaluations of group members uninvolved in the program to determine if intergroup bias could be applied beyond the immediate context. We found that children's intergroup bias toward group members outside their program developed when they were first segregated into classrooms based on their novel groupings and persisted over multiple weeks, adding to our understanding of the impact of categorization on the development and persistence of children's intergroup biases. We consider our findings both in terms of how categorization influences the development of intergroup bias and ways to use re- and cross-categorization to defeat children's intergroup biases. Copyright © 2010 John Wiley & Sons, Ltd.