Higher education policymakers are concerned about the success of first-generation college students. In this study, we investigate one potential factor that may influence outcomes: first-generation students' peers. To mitigate common biases that may arise when estimating peer effects, we leverage the assignment of roommates at The United States Military Academy (West Point). We do not find evidence that being exposed to a roommate(s) with a one standard deviation higher English SAT score impacts first-semester English grades for first-generation students. Our findings for math are inconclusive, with at best suggestive evidence of a small, positive effect.