There is some evidence that occupational therapy as a profession may have adopted supervision without adequate attention to the costs involved in the process, in terms of training, time, commitment and the emotional consequences for the participants (see parts 1 and 2, Sweeney et al 2001a,b). This work suggests that many occupational therapists have This paper, the final one in a series of three, aims to draw together the findings of the two reported studies on the experiences of supervisors and supervisees in the discipline of occupational therapy (Sweeney et al 2001a,b) and to distil recommendations for the future practice of supervision within the profession.A paucity of empirical research on the process of supervision within the health care professions has resulted in a confusion concerning the purpose of supervision, its distinctiveness from other activities and the role and function of both supervisors and supervisees, and a lack of clarity regarding the tasks and techniques to be employed within supervision. In keeping with concerns about supervision expressed in the general literature, the findings of these studies (Sweeney et al 2001a,b) suggest that supervision as it is currently practised within occupational therapy is not a comfortable experience for either the supervisor or the supervisee. This paper recommends steps that can be adopted by both supervisors and supervisees to facilitate the development of a more productive relationship within supervision; for example, the use of training, exposure to theories and models of supervision, and the use of agendas, contracts and feedback. Supervisor and supervisee recommendations are discussed within the wider framework of professional, cultural and organisational implications.