This study examines pedagogical practices in PhD supervision meetings in an Australian university using conversation analysis. Although communication between supervisors and students has been acknowledged as having a great impact on the successful completion of a PhD and several studies of PhD supervision meetings have been conducted, the actual interaction that takes place in this goal-oriented institutional setting is under-researched.Data for this study consist of approximately 25 hours of supervisory talk and video recorded during 25 PhD supervision meetings in an Australian research intensive university. Participants include five supervisors and five PhD students in both social and natural sciences at two stages of candidature: an early stage within the first year of candidature, and a late stage before thesis submission.Analysis of the data shows that supervisors performed a range of actions in accomplishing their challenging task of balancing giving guidance and developing student autonomy. These actions include giving guidance and factual information, giving feedback with equivocation, providing several options, withholding advice, and questioning. In performing these actions and flexibly switching between actions, supervisors sometimes treat the students as needing guidance and advice, and sometimes regard them as independent researchers. The examination of supervisors' pedagogical practices highlights that equivocation is a factor constituting the balancing act.The analysis demonstrates that students are oriented to the goal of learning to become independent researchers. Their orientations centre on the balance between seeking guidance, advice from supervisors and developing their skills and knowledge as independent researchers. The students' orientations towards independent researchers at the early stage of candidature are displayed in the way they take responsibility for their study, voice their opinions, and demonstrate their knowledge and capability of doing research. Meanwhile students at both stages of candidature take responsibility for their study and voice their own opinions, those at the late stage display more confidence and competency. In addition to taking responsibility and voicing opinions, the students at the late stage are in a position to make own decisions on research-related issues and claim independent knowledge.The dynamics of the interaction between the supervisors and students emerge from the analysis of the data and this suggests distinctive features of supervision meetings in the two stages of candidature. As a result, this study provides empirical evidence for an understanding of institutional ii talk in the context of higher education and implications for supervision pedagogy. It extends empirical knowledge of pedagogical practices in supervision meetings. This knowledge will provide supervisors and students with deep insights into their own practices. It can also help to bridge the mismatch of expectations about roles and responsibilities between supervisors and stu...