Theoretical approaches to group therapy and dynamics conveniently depend on what the group is and who is doing the research. Cartwright and Zander gamely argue that many disparate theories merely validate each and propose (with the confidence of their Victorian forebearers) that one day when we know more, a unifying theory arise.

Toseland and Rivas list the following five theories important to group practice:

1. Psychoanalytic theory (Freud)
2. Learning theory (Bandura)
3. Field theory (Lewin)
4. Social exchange, or interaction theory (Blau, Homans, Thibaut, and Kelly)
5. Systems theory (Anderson and Olsen)

Cartwright and Zander add four others:

1. Sociometric orientation
2. General psychology orientation (looking at individual group members)
3. Empiricist-statistical orientation, or group syntality theory
4. Formal models orientation

Therapists must choose between working with the group as a whole and working with individual members in the group. The whole group approach attends to group setting, relationship building, task, and process. According to Yalom, process comments serve to keep the group in the moment and help members appreciate how they relate to one another. This should maintain group awareness and goal orientation. The individual approach is basically individual therapy in a group setting; this approach is sometimes called vertical intervention (as opposed to whole group, horizontal intervention). Kissen suggests that the same therapist may use whole group or the individual approach depending on a group's internal dynamics.

Research into outcome seeks to find out how well groups work, while research into process investigates how they work. What studies there are suggest the heartening result that group therapy is better than no therapy at all, but there's not enough research out there to definitively compare group to other therapy outcomes. From a research standpoint, it's uncertain exactly how and why groups work as treatment.