The Adult Attachment Interview (AAI) was developed by Mary Main and colleagues, and it suggests that attachment-related behaviour can be validly and reliably studied in transcribed interviews and in the interviewee's attentional flexibility and readiness to talk. With the aim to clarify and critique the AAI, we recap the development of the AAI and examine the AAI protocol and the related manual for Adult Attachment Scoring and Classification. We problematise the ways in which the central concepts of “coherence” and “collaboration” are theorised in the AAI on the basis of Grice's theory of conversational implicature and the cooperative principle. We argue that Grice's ideas are used out of context, and that they cannot be used to validly conclude on coherence and collaboration in an interview. The use of Grice's concepts ultimately conceals the subjectivity of the interpreters who can disregard and pathologise appropriate conversational norms from non-Western cultures.