This study aimed to compare involuntary and voluntary clients in the establishment of the therapeutic alliance in the context of family therapy. The system for observing family therapy alliances was used to rate the alliance in sessions 1 and 4 from videotapes of 40 families seen in brief family therapy. This instrument has four alliance dimensions. In the first session, results showed that the clients who sought therapy voluntarily demonstrated more alliance‐related behaviour than did involuntary clients in all alliance dimensions. In the fourth session, however, only the Engagement dimension showed group differences. Notably, there also were group differences in the evolution of the alliance from the first to the fourth session in the Safety dimension, with the voluntary clients developing this dimension more negatively. The results are discussed in terms of the specific characteristics of involuntary clients, as well as the implications for practice. Practitioner points Be aware of differing motives, motivations, and degrees of readiness for therapy within the family Inquire about the amount of pressure experienced by clients, and the source of that pressure, early in the family therapy Promote and monitor the therapeutic alliance with involuntary clients Promote a safe context, providing structure and guidelines for safety and confidentiality and helping clients to talk truthfully