ABSTRACT Maternal sensitivity and disrupted communication are usually considered independently as antecedents of attachment security and attachment disorganization, respectively. This study examined whether considering them jointly allows specific predictions of attachment classifications. The sample (N = 159) was selected from a previous study conducted in Israel between 1991–1993, and over-represented disorganized and ambivalent attachment. Attachment was assessed at 12 months in the Strange Situation Procedure (SSP), sensitivity was assessed from free-play observations at 6 and 12 months, and disrupted communication was coded from the SSP. As hypothesized, high sensitivity and low disruption predicted secure attachment; low sensitivity and high disruption predicted disorganized-insecure attachment or ambivalent attachment; and high sensitivity and high disruption predicted disorganized-secure attachment. Low sensitivity and low disrupted communication did not predict avoidant attachment. The results show that combining maternal sensitivity and disrupted communication improves the precision in identifying maternal antecedents of attachment.