In this study, the author analyzed panel data on elite university students to test prominent explanations for how college attendance affects religious identities and behaviors. Results from random-effects logistic regression showed that in-college religiosity was primarily a function of pre-college religious background. While the secularization hypothesis received no support, aspects of the campus social environment were reliably linked to patterns of religiosity across the college years. Although increasing racial-ethnic diversity in recent decades has coincided with rising aggregate levels of student religiosity on campus, the presence of interracial friendships and roommates predicted significantly lower levels of religious identification and involvement.