Primary among the issues in the study of school-age children with emot
ional and behavioral disorders (EBD) is the prediction and control of
aggressive behavior. Definitional and conceptual problems in measureme
nt have limited our ability to identify and describe classroom conditi
ons that predict occurrences of aggression. In this paper, we revisit
briefly a number of issues in the direct measurement of classroom aggr
ession including low base rates, interactional sequences, and reliabil
ity.