The purpose of the current mixed-methods study was to investigate the effectiveness of a traditional and novel knowledge activation technique for supporting fifth and sixth graders' comprehension of exposition covering unfamiliar content. For the quantitative portion, 149 rural middle-school students were randomly assigned to 1 of 3 conditions: knowledge mobilization (traditional), relational reasoning (novel), or text annotation (control). Readings used during training and for posttraining assessment were chapters from a textbook on ancient Greece and Rome. The posttraining comprehension assessment included multiple-choice locate/recall questions and short-answer integrate/interpret and critique/evaluate questions. The results of structural equation modeling demonstrated the comprehension outcomes for students in the relational reasoning activation condition were significantly higher than for students in either the mobilization ([beta] = 0.41) or control conditions ([beta] = 0.25). Two weeks after training, semistructured interviews were then conducted with 4 high- and 4 low-performing students from the mobilization and relational reasoning training groups. The purpose was to explore students' memory and use of the two activation techniques. Overall, the responses of high-performing students in both training conditions were richer than those for the low-performing students. Also, students in the relational reasoning condition displayed greater facility in the use of that technique than those in the mobilization condition. The contributions of this training study to the research and instruction on the effects of knowledge activation on students' comprehension are delineated and suggestions for subsequent research are forwarded.