Parental pre-natal representations predict the interactive patterns that parents will put in place after childbirth. Early interactions defined by high parental emotional availability (EA) influence the development of security in children. To date, research on the predictive role of parental pre-natal representations on child attachment is still poor. Moreover, investigations on pre-natal representations have mainly focused on mothers. This study aimed at: investigating the criterion validity of the Interview of Maternal Representations During Pregnancy-Revised (IRMAG-R) and of the Interview of Paternal Representations During Pregnancy (IRPAG), using EA, parental attachment, and child attachment toward both parents, as criteria; testing the incremental validity of the IRMAG-R and IRPAG in the prediction of child attachment, controlling for other covariates, such as depressive and anxious levels during pregnancy, EA, and parental attachment; evaluating the possible mediation role of EA on the relationship between parental representations during pregnancy and child attachment. Fifty couples of primiparous parents were recruited during pregnancy, when the IRMAG-R and IRPAG were administered to mothers and fathers. At 6–9 months after childbirth, the mother–child and father–child interactions were coded by means of the EA Scales (EAS). At 14–18 after childbirth, the Adult Attachment Interview (AAI) was administered to parents, and the Strange Situation Procedure (SSP) was carried out to assess children's attachment toward mothers and fathers, respectively. The results showed significant correlations between parental pre-natal representations and EA, parental attachment and child attachment. As regards the prediction of child attachment, the IRMAG-R/IRPAG categories showed: a significant and large unique contribution for maternal representations; a close to be significant contribution for paternal representations (with a higher effect size for mothers than fathers). Moreover, while the indirect effect of pre-natal representations in the prediction of child attachment was not significant for mothers, it was instead significant for fathers. The results of this study confirmed the criterion validity of the IRMAG-R and IRPAG, and supported the incremental validity of the IRMAG-R and IRPAG in the prediction of children's attachment categories. Finally, the mediation models revealed that EA did not mediate the relationship between maternal pre-natal representations and child attachment, while it totally mediated the relationship between paternal pre-natal representations and child attachment.