To reduce bullying, more knowledge on children defending their victimized peers is critical. In previous work, predominantly cross‐sectional in nature, defending has typically been operationalized as one single, broad construct. However, there are good reasons to assume that attacking the bully (bully oriented defending) and comforting the victim (victim‐oriented defending) are relatively independent constructs, with potentially different correlates. This longitudinal study in the Netherlands (N = 394; Mage = 10.3) combined person‐ and variable‐centered techniques to examine relations between two different forms of defending and multiple outcome variables. In addition to the largest group scoring low on both types of defending, three subgroups emerged. A small group of “traditional,” predominantly female defenders, scored high on both forms of defending. These children were well liked and high in reputation‐based status, as indexed by perceived popularity and resource control. A larger, predominantly female second group only scored high on victim‐oriented defending. These children were also well liked, but low in reputation‐based status. The third group only scored high on bully oriented defending, and predominantly contained boys. These children were high in reputation‐based status but quite disliked, and they scored high on bullying. Findings strongly suggest that bully oriented defending does not in all cases reflect desirable interventions of empathic children. Aggr. Behav. 42:585–597, 2016. © 2016 Wiley Periodicals, Inc.