The claim that fathers “swap” families when they form new ones—that is, they shift allegiances from nonresident children to new residential children (e.g., Furstenberg, 1995)—has not been directly evaluated empirically. Drawing on data from the two waves of the National Survey of Families and Households, we test Furstenberg's argument in terms of child‐support transfers to nonresidential children, and we also test an elaboration of his approach that distinguishes between resident biological children and stepchildren. Using static‐score models, our findings indicate that fathers do swap families but only when the trade‐off is between new biological children living inside fathers' households and existing biological children living outside fathers' households. Even though our analytic sample is small, our findings have important implications for child well‐being, child‐support policy, and the meaning of fatherhood.