Despite a burgeoning literature on first-generation college student success, little is known about how the presence of first-generation or continuing-generation peers may facilitate desired outcomes among first-generation students. Moreover, two theoretical traditions lead to contrasting predictions about the potential impact of the representation of other first-generation students. The present study explored this issue within a sample of 4154 first-year, first-generation students who lived on 159 floors in 12 residence halls. Across a variety of statistical models that sought to reduce selection bias, students with a first-generation roommate have lower retention and graduation rates than students with a continuing-generation roommate. Students with a first-generation roommate also have more modest social networks within the university dining halls, but no significant relationships are observed for first-semester or first-year GPA. The link between the proportion of first-generation floormates and success outcomes is rarely significant and yields inconsistent results across analyses.