Despite convincing outcome evidence and the popularity of family therapy, little is about how interpersonal change actually comes about in this context. To synthesize and integrate what is and to offer recommendations for future, clinically relevant research, all published process studies on family therapy were reviewed. Included were naturalistic studies of conjoint treatment in which the focus was the in-session verbal behavior of the participants or their self-reported perceptions of actual interactions. Dating from 1963, 36 studies met the inclusion criteria. The review was organized according to 3 hierarchically ordered levels of in-session processes (L. S. Greenberg, 1986): observations of (a) speech acts, followed by studies of (b) important incidents or change episodes and (c) the therapeutic relationship. The implications of what is known are followed by a discussion of what is not known and suggested directions for future research. Since the 1950s, when family therapy was considered an unusual, if not heretical, mode of treatment, its popularity has been rapidly accelerating. Family therapy is now routinely taught to students in all of the major mental health professions and to practitioners seeking to improve their skills in professional workshops, conferences, and free-standing training centers. The process of family treatment is described, debated, and analyzed not only in practice-oriented periodicals and books but also in wellrespected journals of psychotherapy research. Despite the ever-growing popularity of family therapy, little is about how interpersonal change actually comes about in this context. One reason for this lack of knowledge may be because, to date, investigators have emphasized the global efficacy of family