We describe recent changes in propensities to marry according to the age and educational attainment of potential spouses. Relating actual marriage to the population of men and women at risk enables us to distinguish between changes in the availability of eligible partners and changes in the force of attraction between men and women in particular categories. The procedure is applied to data from 1973, 1980, and 1988 Current Population Surveys. Multivariate analysis suggests that the sharp declines in marriage rates between 1972 and 1979 were not highly differentiated by age or education for either men or women, but that the smaller declines between 1979 and 1987 were highly concentrated among younger women. Age and education homogamy increased during the latter period Including cohabiting unions in the definition of marriage reduces the magnitude of the declines but does not alter their essential patterns.