A central goal of education is to provide students with knowledge they: (a) can successfully apply within the domain of learning and (b) can transfer, as appropriate, to new and different domains. Yet, much research has shown that learners frequently fail to access and use applicable stored knowledge when the circumstances at retrieval differ from those at encoding. Recent laboratory studies have revealed that encoding instances of a concept as members of a relational category considerably attenuates such failures of access. In the present work, we evaluate the efficacy of one such technique (category construction) applied in conjunction with direct instruction in the authentic educational setting of middle-school classrooms. Across two experiments, category construction is evaluated relative to a standard classroom practice of a worksheet with comprehension questions in order to assess how well each promotes mastery and transfer of a target concept. In addition, potential supports to improve learning outcomes are assessed: practice with category construction (Experiment 1) and partial precompletion of the category construction task (Experiment 2). Overall, we find favorable evidence that category construction leads to greater sensitivity to the underlying conceptual structure of targeted concepts and delivers transfer advantages over a worksheet control. (PsycInfo Database Record (c) 2024 APA, all rights reserved)