Bowlby proposed that the individual's social experiences, as early as in infancy, contribute to the construction of Internal Working Models (IWMs) of attachment, which will later guide the individual's expectations and behaviors in close relationships all along his or her life. The qualitative, individual characteristics of these models reflect the specificity of the individual's early experiences with attachment figures. The attachment literature globally shows that the qualities of IWMs are neither gender specific nor cultural specific. Procedures to evaluate IWMs in adulthood have been well established, based on narrative accounts of childhood experiences. Narrative procedures at earlier ages (e.g., in the preschool years) have been proposed, such as Bretherton's Attachment Story Completion Task (ASCT), to evaluate attachment representations. More than 500 ASCT narratives of preschoolers, coming from five different countries, have been collected, in the perspective of examining possible interactions between gender and culture regarding attachment representations. A specific Q-Sort coding procedure (CCH) has been used to evaluate several dimensions of the narratives. Girls' narratives appeared as systematically more secure than those of same-age boys, whatever their culture. The magnitude of gender differences, however, varied between countries. Taylor's model of gender-specific responses to stress and Harwood's and Posada's hypothesis on inter-cultural differences regarding caregiving are evoked to understand the differences across gender and countries.