<p>How should colleges assign students to academic advisors? I address this question using administrative data from a small Dutch liberal arts college in which faculty advisors are randomly assigned to students. I estimate how advisor–student matching on three characteristics, namely academic department, origin, and gender, affects student grades, academic standing, and exit. I find that matching on academic department has beneficial effects, whereas matching on origin and gender does not. I interpret these results through the similarity-attraction paradigm, treating department, origin, and gender as proxies for deep-level, experiential, and surface-level similarity, respectively. The findings suggest that disciplinary proximity is a more important dimension of advisor–student similarity than shared demographic background and that advisor assignment can be used as a low-cost tool to improve student success.</p>