Attachment theory promoted an impressive body of research on the psychic developmental processes, resulting in studies on both typical and atypical development. Much of the diffusion of the attachment theory in the clinical field was related to the design of reliable instruments to evaluate the organization of attachment in infancy as well as in adulthood. Until recently, the lack of a suitable instrument to assess attachment in middle childhood as well as in adolescence hindered the expansion of research in these developmental phases during which the parent-child relationship takes on a different, albeit still crucial, role. The Child Attachment Interview (CAI), a measure that was recently designed to assess attachment at a representational level in middle childhood and adolescence, filled the measurement gap. The aim of the current review was to summarize previous empirical investigations concerning CAI in order to (a) provide an overview of the state of current research, (b) identify unanswered questions, and (c) propose future research directions. A narrative review was conducted to map the current research findings by searching for the term “Child Attachment Interview” in the Medline, Scopus, Web of Science, and PsychINFO databases, followed by a search in Mendeley. Limits were set to exclude dissertations, chapters in books, and qualitative or theoretical papers, while empirical studies were included if they used the CAI and were published in English language, peer-reviewed journals by July, 2016. The review, which ultimately included 39 studies meeting the criteria, showed that the CAI is a reliable instrument to assess attachment organization in clinical and non-clinical samples, thus providing a worthwhile contribution to the investigation of the influence of the parent-child relationship beyond infancy and early childhood. Nevertheless, the review pointed out a number of relevant open issues, the most critical of which concerned the CAI coding and classification system. In particular, some relevant questions arose about (a) how opportune it would be to maintain a distinct classification for mother and father, (b) coding challenges regarding both the father and the Preoccupied and Disorganized classification, and finally (c) the advantage of a dimensional vs. a categorical approach.