The prototype hypothesis suggests that attachment representations derived in infancy continue to influence subsequent relationships over the life span, including those formed with one's own children. In the current study, we test the prototype hypothesis by exploring (a) whether child-specific representations following actual experience in interaction with a specific child impacts caregiver-child attachment over and above the prenatal forecast of that representation and (b) whether maternal attachment representations exert their influence on infant attachment via the more child-specific representation of that relationship. In a longitudinal study of 84 mother-infant dyads, mothers' representations of their attachment history were obtained prenatally with the Adult Attachment Interview (AAI; M. Main, R. Goldwyn, & E. Hesse, 2002), representations of relationship with a specific child were assessed with the Working Model of the Child Interview (WMCI; C.H. Zeanah, D. Benoit, & L. Barton, 1986), collected both prenatally and again at infant age 11 months, and infant attachment was assessed in the Strange Situation Procedure (M.D.S. Ainsworth, M. C. Blehar, E. Walters, & S. Wall, 1978) when infants were 11 months of age. Consistent with the prototype hypothesis, considerable correspondence was found between mothers' AAI and WMCI classifications. A mediation analysis showed that WMCI fully accounted for the association between AAI and infant attachment. Postnatal WMCI measured at 11 months' postpartum did not add to the prediction of infant attachment, over and above that explained by the prenatal WMCI. Implications for these findings are discussed.