This longitudinal study of forty-four families explored fathers’ as compared to mothers’ specific contribution to their children's attachment representation at ages 6, 10, and 16 years. In toddlerhood, fathers’ and mothers’ play sensitivity was evaluated with a new assessment, the sensitive and challenging interactive play scale (SCIP). Fathers’ SCIP scores were predicted by fathers’ caregiving quality during the first year, were highly consistent across 4 years, and were closely linked to the fathers’ own internal working model of attachment. Qualities of attachment as assessed in the Strange Situation to both parents were antecedents for children's attachment security in the Separation Anxiety Test at age 6. Fathers’ play sensitivity and infant–mother quality of attachment predicted children's internal working model of attachment at age 10, but not vice versa. Dimensions of adolescents’ attachment representations were predicted by fathers’ play sensitivity only. The results confirmed our main assumption that fathers’ play sensitivity is a better predictor of the child's long-term attachment representation than the early infant–father security of attachment. The ecological validity of measuring fathers’ sensitive and challenging interactive play behavior as compared to infant proximity seeking in times of distress is highlighted. Findings are discussed with respect to a wider view on attachment in that both parents shape their children's psychological security but each in his or her unique way.