Using retrospective reporting as a source of “longitudinal” data, an analysis is presented of the varying degrees of homogamy between a homogeneous sample of engaged girls and three men with whom each had had a serious relationship. Evidence is presented which suggests that (a) there is a general tendency to become more homogamous through the mate selection period on some dimensions (e.g., education) but to become less so on other dimensions (e.g., urbanity); (b) some of the patterns more easily fit a theory based on variations in opportunities to meet different kinds of partners, others more easily fit a normative theory; (c) social changes during the selection period (geographical or social mobility) influence the level and pattern of homogamy.