Background: Unpaid carers are the mainstay of support for people with dementia. Admiral Nursing (AN) is the only specialist nursing service that specifically focuses on supporting such carers, but evidence of its effectiveness, costs and relationships with other health and social care services is limited. This project aimed to address this gap and explore the feasibility of a full-scale formal evaluation. Objectives: To explore the relationships between characteristics of carers and people with dementia, service type and input and outcomes; to develop and test data collection methods for subsequent economic evaluation; to explore the effect of AN on outcomes and costs, compared with usual care; to explore the perceived system-wide impact of specialist support for carers of people with dementia, compared with usual care; and to implement new data collection methods in AN, which could also be used by other services, to facilitate evaluation. Design: A mixed-methods study, using secondary analysis of an administrative data set, and primary (cross-sectional) quantitative and qualitative data collection. Setting: Qualitative research with carers in four areas of England; a survey of carers in 32 local authority areas (16 with and 16 without AN); and qualitative interviews with professionals in four areas. Participants: Thirty-five carers of people with dementia and 20 professionals were interviewed qualitatively; 346 carers completed in-scope questionnaires (46% through AN services and 54% from matched non-AN areas). Interventions: Specialist nursing support for carers of people with dementia (with AN as an exemplar) compared with usual care. Main outcome measures: The Adult Social Care Outcomes Toolkit for Carers; the EuroQol-5 Dimensions, five-level version; and the Caregiver Self-Efficacy for Managing Dementia Scale. Data sources: Dementia UK’s AN administrative data set. Results: Admiral Nurses are successfully targeting the most complex cases. They work predominantly with older carers who have the main responsibility for the person with dementia, who are heavily involved in caring activity and who may be at risk. Three outcome areas that are important to carers of people with dementia and are potentially affected by receiving support are (1) carer self-efficacy, (2) carer quality of life (3) and carer mental and physical health. The carers in the survey receiving support from AN were older, were more heavily involved in caring and had poorer outcomes than carers not in receipt of such support. When these differences were controlled for, carers supported by AN had better outcomes, although the differences did not reach statistical significance. Health and social care costs were similar in both groups. The perceived system-wide impact of services, such as AN, is not well understood by professional stakeholders. Limitations: Challenges were experienced in identifying similar carers in areas with or without an AN service and in the cross-sectional nature of the work. Conclusions: Specialist nursing support to carers of people with dementia may enable them to continue providing care to the end or very close to the end of the dementia journey. The outcomes for such carers may be no different from, or even slightly better than, those of similar carers without this support, although the costs to health and social care services are the same in each case. Future work: Future research could investigate the impact of specialist support for carers on admission to long-term care. There is also a need for more work to encourage routine use of the selected outcome measures in dementia service delivery. Funding: The National Institute for Health Research Health Services and Delivery Research programme.