Background: In the UK, most people with dementia die in the community and they often receive poorer end-of-life care than people with cancer.; Objective: The overall aim of this programme was to support professionals to deliver good-quality, community-based care towards, and at, the end of life for people living with dementia and their families.; Design: The Supporting Excellence in End-of-life care in Dementia (SEED) programme comprised six interlinked workstreams. Workstream 1 examined existing guidance and outcome measures using systematic reviews, identified good practice through a national e-survey and explored outcomes of end-of-life care valued by people with dementia and family carers ( n = 57) using a Q-sort study. Workstream 2 explored good-quality end-of-life care in dementia from the perspectives of a range of stakeholders using qualitative methods (119 interviews, 12 focus groups and 256 observation hours). Using data from workstreams 1 and 2, workstream 3 used co-design methods with key stakeholders to develop the SEED intervention. Worksteam 4 was a pilot study of the SEED intervention with an embedded process evaluation. Using a cluster design, we assessed the feasibility and acceptability of recruitment and retention, outcome measures and our intervention. Four general practices were recruited in North East England: two were allocated to the intervention and two provided usual care. Patient recruitment was via general practitioner dementia registers. Outcome data were collected at baseline, 4, 8 and 12 months. Workstream 5 involved economic modelling studies that assessed the potential value of the SEED intervention using a contingent valuation survey of the general public ( n = 1002). These data informed an economic decision model to explore how the SEED intervention might influence care. Results of the model were presented in terms of the costs and consequences (e.g. hospitalisations) and, using the contingent valuation data, a cost–benefit analysis. Workstream 6 examined commissioning of end-of-life care in dementia through a narrative review of policy and practice literature, combined with indepth interviews with a national sample of service commissioners ( n = 20).; Setting: The workstream 1 survey and workstream 2 included services throughout England. The workstream 1 Q-sort study and workstream 4 pilot trial took place in North East England. For workstream 4, four general practices were recruited; two received the intervention and two provided usual care.; Results: Currently, dementia care and end-of-life care are commissioned separately, with commissioners receiving little formal guidance and training. Examples of good practice rely on non-recurrent funding and leadership from an interested clinician. Seven key components are required for good end-of-life care in dementia: timely planning discussions, recognising end of life and providing supportive care, co-ordinating care, effective working with primary care, managing hospitalisation, continuing care after death, and valuing staff and ongoing learning. Using co-design methods and the theory of change, the seven components were operationalised as a primary care-based, dementia nurse specialist intervention, with a care resource kit to help the dementia nurse specialist improve the knowledge of family and professional carers. The SEED intervention proved feasible and acceptable to all stakeholders, and being located in the general practice was considered beneficial. None of the outcome measures was suitable as the primary outcome for a future trial. The contingent valuation showed that the SEED intervention was valued, with a wider package of care valued more than selected features in isolation. The SEED intervention is unlikely to reduce costs, but this may be offset by the value placed on the SEED intervention by the general public.; Limitations: The biggest challenge to the successful delivery and completion of this research programme was translating the ‘theoretical’ complex intervention into practice in an ever-changing policy and service landscape at national and local levels. A major limitation for a future trial is the lack of a valid and relevant primary outcome measure to evaluate the effectiveness of a complex intervention that influences outcomes for both individuals and systems.; Conclusions: Although the dementia nurse specialist intervention was acceptable, feasible and integrated well with existing care, it is unlikely to reduce costs of care; however, it was highly valued by all stakeholders (professionals, people with dementia and their families) and has the potential to influence outcomes at both an individual and a systems level.; Future Work: There is no plan to progress to a full randomised controlled trial of the SEED intervention in its current form. In view of new National Institute for Health and Care Excellence dementia guidance, which now recommends a care co-ordinator for all people with dementia, the feasibility of providing the SEED intervention throughout the illness trajectory should be explored. Appropriate outcome measures to evaluate the effectiveness of such a complex intervention are needed urgently.