Behavioral interventions have demonstrated clinical utility in improving the behavior of children with ADHD, especially in specialized therapeutic milieus (Pelham et al., 2000). Improvements in children's target behaviors often occur in the treatment settings where contingencies are in place and delivered consistently. However, generalization of treatment effects across settings and over time--the overarching clinical objective of psychosocial interventions-remains an elusive goal. Although programmed efforts to enhance generalization have been incorporated into numerous ADHD treatment studies (e.g., Abikoff et al., 2004; Pfiffner & McBurnett, 1997; Wells et al., 2000), in non-treatment settings and when treatment has ended, children with ADHD often have difficulties behaving in ways appropriate to the situation and setting (Hinshaw, Klein, & Abikoff, 2002; Pfiffner, Calzada, & McBurnett, 2000). A question rarely posed is "Why is it so difficult to achieve generalization?" In addressing this issue, it is important to consider, at a basic level, what is required for generalization to take place. First, a behavior must be learned before it can be expected to occur consistently in multiple settings and over time. Second, the child must be able to identify numerous situations, all of which are relevant exemplars that call for the learned behavior, even though elements of these situations differ. Third, the child must be able to modify learned behaviors appropriately in response to particular contexts and situational demands. This essay is intended to refocus attention on the topic of generalization, which for more than 30 years has remained a major objective in ADHD psychosocial clinical research. The hope is that this commentary will lead to studies that examine this issue in depth in individuals with ADHD. Information from this line of research could potentially broaden the understanding of the disorder, enhance treatment development research, and lead to intervention strategies that have broad clinical impact.