National data for ever-married men age 20 to 65 in March 1973 are utilized to estimate least squares and log-linear structural equation models of age at marriage. We demonstrate that most characteristics of family background (including both the family structure and its socioeconomic standing) are irrelevant in their effect on age at marriage. Intercohort trends are not explicable with reference to the changing socioeconomic, ethnic, or nativity compositions of the cohorts. Regional differences in age at marriage have persisted over the years in only slightly diminished form and cannot be explained by reference to the nativity, ethnic, or socioeconomic compositions of the regions. Early job status relates only weakly to age at marriage. Only those activities that are time-consuming or otherwise disruptive of the smooth operation of normal life-cycle processes during the transition from adolescence to adulthood (such as college attendance and service in the military) seriously affect the age at which a man marries.