In their pioneering research, Becker, Landes and Michael (1977) found that beyond age 30 there is a positive relationship between women's age at first marriage and marital instability. They interpreted this finding as a "poor-match" effect emerging as the biological clock begins to tick. In analyses of the 2006-2010 National Surveys of Family Growth (NSFG), we find evidence of the existence of this effect: women who delay marriage disproportionately make unconventional matches, which are generally associated with high marital instability (N = 3,184). We also find, however, that their unions are very solid. We develop and test competing hypotheses that can account for these patterns. In addition, noting that women's delayed transition to first marriage has been accompanied by higher proportions of women entering marriage with 16 years of schooling or more, we examine changes across the last three NSFG cycles in the education - marital instability association.