This thesis aims to contribute to the understanding of research higher degree supervision and thereby its enhancement. My study departs from the current emphasis on issues of practice to offer a set of scholarly understandings embedded in the beliefs that supervisors and candidates hold about supervision and closely related academic mailers. It is aligned with the movement over the past two decades towards concentrating on understanding why teachers and students behave in particular ways, rather than describing what they do and how they do it. I draw on the literature of research higher degree (RI-ID) supervision, the conceptual framework of beliefs research and Gadamer's concept of the hermeneutic circle to argue that supervision is best understood as a plexus of closely related educational beliefs about research, teaching, learning and supervision. Research from this perspective acknowledges supervisions plural, multifunctional character and its holistic nature. The beliefs construct recognises the powerful effect that individualsibeliefs and attitudes have on the way they define educational tasks, make related decisions and prefer to act. Thirty-four participants from one engineering faculty were interviewed about their beliefs about the four components of the supervision plexus. Entire transcripts were coded, using a three-phase, inductive method of analysis incorporating constant-comparative techniques and conceptual field principles, to reveal individuals integrated thinking about the whole process of supervision. This method ensures that the findings remain embedded in the data and retain the richness of individual experience. I identified four different core tendencies to the plexus, based on two bipolar frames - controlling/guiding and task-focussed/person-focussed kinds of beliefs. The result is four global orientations to supervision: controlling/task-focussed, controlling/person-focussed, guiding/task-focussed and guiding/person-focussed. Subcategories accommodate individuals whose beliefs differ in specific aspects but whose focal beliefs fit the global group. Each orientation is elaborated by an orientation belief profile - an integrated system of beliefs about the aspects of the plexus that are common to the individuals in that category iso the profiles describe the orientations as much as they describe the individuals in each category. The beliefs in each profile are organised into six belief clusters and different dimensions of the beliefs describe each orientation. To show the location, density and type of inter-linkages among beliefs and belief clusters orientation webs were drawn. The four webs exhibit a high degree of interconnectedness among beliefs, confirming my contention of a supervision plexus of co-dependent and logically interrelated components. Research findings indicate that practitioners beliefs about teaching are central and powerful in determining their supervisory goals and their predisposition towards particular pedagogical approaches to achieving them. With this advanced understanding of the pedagogy of supervision, a case is built for viewing research higher degree supervision as a teaching activity within the university, and positing its management as a 'joint portfolio' between the teaching and learning centre and the research centre of the university. Other findings are that controlling/task-focussed beliefs are generally favoured by RI-ID candidates and that guiding/person-focussed beliefs more commonly describe the way supervisors think about supervision. Although their strategic enactment may differ according to circumstance, beliefs were found to be consistent across contexts. The supervisors role in shaping candidates' beliefs is seen to be diminished by the influence of candidates' preexisting beliefs about teaching. The study establishes a variety of understandings about supervision within this one engineering faculty, suggesting that pedagogical understandings may be more powerful than disciplinary expectations and attitudes as determinants of supervisory behaviour. The view of RI-ID supervision discussed in this thesis builds on the earlier research in meaningful ways that enhance our understanding of the process as a whole. The thesis provides possibilities for linking that research with more fruitful and rewarding doctoral experiences for supervisors and candidates.