This part-reflective, part-conceptual article focuses on the issue of supervision models and the need to move away from the dominance of the traditional master-apprentice model which continues to persist in many social sciences and humanities contexts. Drawing on communities of practice theory, I critically reflect on my own PhD supervision experience in Australia in relation to my current South African context in order to highlight how varying forms of a group model of supervision can enhance the doctoral experience by creating more opportunities for PhD candidates to engage in social learning. Furthermore, the article discusses some of the challenges associated with maintaining group coherence in a community of practice due to the fundamentally social nature of knowledge and the associated contestations around the legitimacy of knowledge and ways of knowing and the effect this has on a community. The article concludes with a suggestion of how this challenge can be mitigated through the creation of smaller, theory-based research 'clusters' within larger doctoral programmes.