We develop a model of college admissions with interdependent preferences and noisy signals, building on Milgrom & Weber (1982) and Azevedo & Leshno (2016). Using this model, we evaluate the equilibrium consequences of making admissions more correlated across schools. We show that increased correlation makes applicants better off, particularly those who are weaker, and makes matching less assortative. Our analysis also offers an informational rationale for why acceptance rates for early admissions exceed those of regular admissions.