Background: Aphasia language therapy sessions have a general “order of phases” and usually open with a period of casual conversation. It has been noted that the overall structure of participation in therapy, characterised by therapist control, is anticipated in this opening phase. The detailed mechanisms through which control is achieved and sustained have not been fully examined. The processes through which these mechanisms operate and their relationship with participant roles and identities appear to be especially relevant for study if the overall goal of aphasia therapy is to address life participation and the social context of communication (e.g., Shadden & Agan, 2004). Aims: The concern of this paper is to examine in detail some of the mechanisms through which professional control is achieved and sustained, with a particular focus on ways in which speech and language therapists and people with aphasia work at generating and maintaining the topics that are observed to develop in the opening phase of aphasia language therapy sessions. In addition, there is a specific interest in how, in certain instances, the activity of topic generation and maintenance may be shown to relate to the participants' identities as “professional” and “patient”. Methods & Procedures: The paper focuses on analysis of data from the opening phase of five sessions, one each from three therapist/person with aphasia dyads, and two from separate sessions of a fourth dyad. Data from these sessions are taken from a larger study of aphasia language therapy in day‐to‐day practice. The turn‐by‐turn process of topic generation is examined in detail using techniques from Conversation Analysis in conjunction with a consideration of “membership categorisation” and “category‐bounded activities”. Outcomes & Results: Close examination of the process of topic generation and maintenance in five sessions from four different dyads revealed how therapists selected certain topics for topicalisation and rejected others. Selected topics were related to activities closely tied to the person with aphasia's identity as a person with communication impairment. In so doing the therapists established a particular type of therapeutic identity for themselves. Conclusions: Managing the person with aphasia's identity by addressing communication activities that are bound to the aphasic impairment and rejecting other types may, perversely, hamper a speech and language therapist's understanding of that person with aphasia's communicative competence, including broader issues that may be germane to intervention. Understanding and recognising where control resides and through which mechanisms it operates is of potential benefit to therapists and people with aphasia alike—for example, enabling more focused and realistic goal setting or more accurate insights into the impact of aphasia, and aphasia language therapy on everyday social participation. The challenge for therapists and for people with aphasia is to be able to shift roles appropriately and act as equals in the enterprise of therapy.