Perhaps the greatest value of response to intervention (RTI) as a decision framework is that it brings attention to variables (e.g., mastery of prerequisite skills, frequency of instructional corrective feedback, reinforcement schedules for correct responding) that if changed might make a meaningful difference for students (e.g., child rate of learning is accelerated and child learns to read). Yet, RTI is also a model that has diagnostic implications. Classification agreement analyses can quantify accuracy and efficiency of decisions within RTI systems and can indicate which procedures and decision rules lead to the best decisions. Because RTI systems are dynamic and procedures vary between schools and implementers, classification agreement analyses, when not misapplied or misinterpreted, are ideal for quantifying utility of various decision models. This article offers a critical analysis of sensitivity, specificity, positive predictive power, and negative predictive power estimates and discusses the development of sophisticated models of decision making.