OBJECTIVES: In the UK policy recommends that service users (patients, carers and the public) should be involved in all publicly funded health and social care research. However, little is known about which approaches work best in different research contexts and why. The purpose of this paper is to explain some of the theoretical limitations to current understandings of service user involvement and to provide some suggestions for theory and methods development. This paper draws upon findings from a review of the research 'evidence' and current practice on service user involvement in the design and undertaking of nursing, midwifery and health visiting research. DESIGN: A multi-method review was commissioned by the NHS Service Delivery and Organisation (SDO) Research and Development Programme. The timeframe was April 2004-March 2005. The full report (Ref: SDO/69/2003) and supplementary bibliography are available from: http://www.sdo.lshtm.ac.uk. REVIEW METHODS/DATA: Initial searches of the health and social care literature and consultations with researchers were used to develop a broad definition of the topic area. A service user reference group (26 members) worked with the project team to refine the scope of the review, to set inclusion criteria and develop a framework for the analysis. Systematic searches of the literature were undertaken online and through library stacks (345 relevant documents were identified). Ongoing and recently completed studies that had involved service users were identified through online databases (34 studies) and through a national consultation exercise (17 studies). Selected studies were followed up using telephone interviews (n=11). Members of the service user reference group worked with the research team to advise on key messages for dissemination to different audiences. RESULTS: Information was gained about contextual factors, drivers, concepts, approaches and outcomes of service user involvement in nursing, midwifery and health visiting research, as well as developments in other research fields. Synthesis of this information shows that there are different purposes and domains for user involvement, either as part of researcher-led or user-led research, or as part of a partnership approach. A number of issues were identified as being important for future research. These include: linking different reasons for service user involvement with different outcomes; understanding the relationship between research data and service user involvement, and developing conceptualisations of user involvement that are capable of accommodating complex research relationships. Suggestions for the development of practice include: consideration of diversity, communication, ethical issues, working relationships, finances, education and training. CONCLUSIONS: Because research is undertaken for different reasons and in different contexts, it is not possible to say that involving service users will, or should, always be undertaken in the same way to achieve the same benefits. At a research project level uniqueness of purpose is a defining characteristic and strength of service user involvement.