The authors use nationally representative survey data and a research design that relies on contemporaneous within-student and within-teacher comparisons across two academic subjects to estimate how class size affects certain non-cognitive skills in middle school. Their results indicate that smaller eighth-grade classes are associated with improvements in several measures of school engagement, with effect sizes ranging from 0.05 to 0.09 and smaller effects persisting 2 years later. Patterns of selection on observed traits and falsification exercises suggest that these results accurately identify (or possibly understate) the causal effects of smaller classes. Given the estimated earnings impact of these non-cognitive skills, the implied internal rate of return from an eighth-grade class-size reduction is 4.6% overall, but 7.9% in urban schools. (Contains 15 notes, 10 tables, and 2 figures.)