Attachment theorists assume that maternal mental representations influence responsivity, which influences infant attachment security. However, primary studies do not support this mediation model. The authors tested mediation using 2 mother-infant samples and found no evidence of mediation. Therefore, the authors explored sensitivity as a moderator, studying the (a) interaction of mental representation and sensitivity as it predicts infant attachment security and (b) level of sensitivity in mothers whose infants' attachment security is either concordant or discordant with their own. The interactional analyses were not significant. But the match-mismatch data showed that when mother-infant attachment strategies were discordant, maternal sensitivity was more consistent with infant than maternal attachment strategy. These findings are congruent with an interpretation of sensitivity as a moderator that can block transmission of attachment strategy.