Abstract Background The move towards enhancing teamwork and interprofessional collaboration in health care raises issues regarding the management of professional boundaries and the relationship among health care providers. This qualitative study explores how roles are constructed within interprofessional health care teams. It focuses on elucidating the different types of role boundaries, the influences on role construction and the implications for professionals and patients. Methods A comparative case study was conducted to examine the dynamics of role construction on two interprofessional primary health care teams. The data collection included interviews and non-participant observation of team meetings. Thematic content analysis was used to code and analyze the data and a conceptual model was developed to represent the emergent findings. Results The findings indicate that role boundaries can be organized around interprofessional interactions (giving rise to autonomous or collaborative roles ) as well as the distribution of tasks (giving rise to interchangeable or differentiated roles ). Different influences on role construction were identified. They are categorized as structural (characteristics of the workplace), interpersonal (dynamics between team members such as trust and leadership) and individual dynamics (personal attributes). The implications of role construction were found to include professional satisfaction and more favourable wait times for patients. A model that integrates these different elements was developed. Conclusions Based on the results of this study, we argue that autonomy may be an important element of interprofessional team functioning. Counter-intuitive as this may sound, we found that empowering team members to develop autonomy can enhance collaborative interactions. We also argue that while more interchangeable roles could help to lessen the workloads of team members, they could also increase the potential for power struggles because the roles of various professions would become less differentiated. We consider the conceptual and practical implications of our findings and we address the transferability of our model to other interprofessional teams.