The promotion of marriage and two-parent families became an explicit public policy goal with the passage of the 1996 welfare reform bill. Marriage has the putative effect of reducing welfare dependency among single mothers, but only if they marry men with earnings sufficient to lift them and their children out of poverty. Newly released data from the 2002 cycle of the National Survey of Family Growth (NSFG), along with data from the 1995 cycle, allow us to compare pre- and post-PRWORA differences in (1) cumulative marriage rates among unwed mothers, and (2) patterns of marital choice (that is, differences in characteristics of the men these mothers marry, such as their education and employment status). Overall, our results show that unwed childbearing is associated with lower marriage rates and marital quality. Difference-in-difference models show that welfare reform was not strongly associated with pre- and post-welfare reform changes in marriage among nonmarital birth mothers, even among the most disadvantaged mothers. Compared with other women, nonmarital birth mothers also were less likely than other women to marry "economically attractive" men in the post-welfare reform period. The success of marriage promotion initiatives may depend heavily on whether women themselves are "marriageable" and whether potential spouses have the ability to support a stable family life. (Contains 1 figure, 5 tables and 8 footnotes.)