Research on personal relationships inescapably uses taxonomies for classifying people, relationships, and interpersonal processes and events, and often develops new ones. However, constructing a good taxonomy is no simple matter. Many existing classification methods have serious weaknesses, and they carry the risk of drawing taxonomic distinctions that are spurious. In addition, researchers frequently make unwarranted assumptions about the nature of the taxonomic categories that they employ in their work. This article introduces a family of quantitative methods for testing and generating taxonomies. Although they have seen little use to date outside of psychopathology, personality psychology, and behavioral genetics, these methods are versatile and readily adapted to the domain of personal relationships, where they offer many research possibilities. Three of the methods are illustrated in a study of elementary forms of relationships.