This chapter considers matching by categories, beginning with a discussion of a specific but empirically very relevant family of models in order to provide a richer representation of heterogeneity between individuals that would account not only for economic aspects such as income or education, as well as more subjective (and less easily observable) ones, such as idiosyncratic preferences for marriage in general and for specific types of spouses in particular. The chapter explores a simple model that describes matching on income or education, the separability assumption, how separability can be justified, and the dual structure under separability. It also provides an overview of the Choo-Siow model, focusing on its basic structure, the matching function, heteroskedasticity, comparative statics, testability and identifiability, and the Galichon and Salanié's cupid framework as an extension of the model.