A high level of clustering distinguishes social networks from other types of networks. Homophily, or the tendency for similar individuals to befriend each other, is commonly purported to generate this phenomenon. While much extant research on homophily emphasizes factors that structurally induce similarity among individuals, less work has investigated how cultural tastes result in individuals preferentially selecting homogeneous alters and how best to approach this methodologically. Using novel data, this study employs statistical network analyses to infer how music tastes produce different structures within an emerging social network by influencing the number and similarity of friends that a person has. I find music tastes vary in the extent to which they produce homophily but that, in order to uncover these dynamics, one must account for the number of people with whom an individual interacts. I conclude with a theoretical discussion of the findings and suggest future research directions.