Although many agree that writing can make a unique contribution to learning, harnessing that contribution is difficult, particularly in the middle grades. The purpose of this study was to measure the efficacy of a feasible writing-to-learn intervention in Grade 6-11 science classes. We focused on middle school students because this group has been least responsive to writing interventions in previous research. We conducted an 8-week intervention in which students completed short metacognitive and longer argumentative writing tasks. Our results indicated that overall students' ability to engage in scientific rhetoric improved, and we identified minimal difference in the growth of middle and high school students. We therefore conducted a cluster analysis to examine how learner-level variables impacted the effectiveness of the intervention. Most notable was that students with low writing motivation and science knowledge showed the largest response to the intervention. These findings suggest that writing-to-learn can be implemented in secondary science to support scientific literacy development.