Data were collected about 95 fourth-grade and fifth-grade children from three sources (teachers, parents, and children) on four constructs related to aggression (proactive aggression, reactive aggression, callous–unemotional traits, and anger dysregulation) by using a multitrait–multimethod design. Two goals were addressed through a correlated-trait–correlated-method analysis: (a) Investigate the overlap versus distinctiveness of proactive aggression, reactive aggression, callous–unemotional traits, and anger dysregulation by examining the correlations among their trait factors. (b) Investigate the extent to which the informant who assessed children’s aggression was associated with the rating given. Moderate-to-strong trait factor correlations emerged among all pairs of latent trait factors, suggesting at best modest discriminant validity among the constructs. The informant played a significant role for almost all ratings; however, teachers’ ratings were the most strongly linked to method variance, and children’s ratings provided the most convergent validity with other sources. Findings are discussed in terms of enhanced assessment of childhood aggression.