We study one aspect of the link between welfare and unwed motherhood: the relationship between benefit levels and the time-to-first-marriage and time-to-next-birth among women whose first" child was born out of wedlock. We use twin births to generate effectively random variation in welfare benefits among mothers within a state, which allows us to control for unobservable characteristics of states that typically confound the relationship between welfare payments and behavior. The twins approach yields evidence that higher base levels of welfare benefits: (1) lead initially unwed white mothers to forestall their eventual marriage; and (2) lead initially unwed black mothers to hasten their next birth. The magnitudes of these effects are small, however. Moreover, we find no evidence that the incremental benefit paid upon the birth of an additional child affects fertility.