Although the common wisdom — based largely on Vital Statistics available only for the period through 1988 — is that marriage is fast becoming a defunct institution, evidence derived from the annual March CPS (1964-95) is used to demonstrate that there have been substantial changes in the long-term trends in marriage and divorce within the 20-24 age group since 1988. For males and females working fulltime year round, and for those with less than a completed college education, age-specific marriage rates appear to have stabilized — and divorce rates have even reversed their direction. This change in a longstanding trend appears to be a cohort-related, rather than a period, phenomenon, since it clearly began with the youngest age groups and is gradually spreading to older groups over time. In addition, the patterns show a clear relationship with male relative income, as hypothesized by Richard Easterlin. However, the relationship between male relative income and relative cohort size appears to have weakened substantially in recent years, and a 'marriage squeeze' variable based directly on relative cohort size produces counterintuitive signs on a number of variables in a full model, and loses its significance when entered along with a relative income variable. Thus it is not clear that marriage and divorce rates can be forecast directly on the basis of relative cohort size: their relationship appears to be closest with male relative income.