Background Case management has been suggested as a way of improving the quality and cost-effectiveness of support for people with dementia. In this study we adapted and implemented a successful United States’ model of case management in primary care in England. The results are reported elsewhere, but a key finding was that little case management took place. This paper reports the findings of the process evaluation which used Normalization Process Theory to understand the barriers to implementation. Methods Ethnographic methods were used to explore the views and experiences of case management. Interviews with 49 stakeholders (patients, carers, case managers, health and social care professionals) were supplemented with observation of case managers during meetings and initial assessments with patients. Transcripts and field notes were analysed initially using the constant comparative approach and emerging themes were then mapped onto the framework of Normalization Process Theory. Results The primary focus during implementation was on the case managers as isolated individuals, with little attention being paid to the social or organizational context within which they worked. Barriers relating to each of the four main constructs of Normalization Process Theory were identified, with a lack of clarity over the scope and boundaries of the intervention (coherence); variable investment in the intervention (cognitive participation); a lack of resources, skills and training to deliver case management (collective action); and limited reflection and feedback on the case manager role (reflexive monitoring). Conclusions Despite the intuitive appeal of case management to all stakeholders, there were multiple barriers to implementation in primary care in England including: difficulties in embedding case managers within existing well-established community networks; the challenges of protecting time for case management; and case managers’ inability to identify, and act on, emerging patient and carer needs (an essential, but previously unrecognised, training need). In the light of these barriers it is unclear whether primary care is the most appropriate setting for case management in England. The process evaluation highlights key aspects of implementation and training to be addressed in future studies of case management for dementia.