Students in a gateway college science course were assigned to two instructional conditions: explanation-only or augmented instruction. In the explanation-only condition, students wrote explanations for six excerpts from their textbook. The augmented-instruction condition wrote explanations on only four of the excerpts, with the other two serving as examples for a lesson about the rhetorical structure of the texts and how to integrate multiple text ideas to answer inference test questions. Both conditions showed pre-post gains in students' metacomprehension skills. In the explanation-only condition, students at-risk for failure (first-generation college, ESL, low income, low ACT) performed significantly worse than other students on course exams, but this difference was reduced by half with the lessons included in the augmented condition.