Action Research (AR) is promoted for health care development. A systematic review was undertaken to gain insight into the uptake and designs of practice-based AR. Empirical research papers from 2000 to 2005 were extracted from CINAHL, MEDLINE and British Nursing Index, and two specialist AR journals. The initial search identified 335 papers: 38% were AR (20% were phenomenology; 32% ethnography; 10% randomised-controlled trials). Further filtering produced 62 AR papers for detailed analysis. Eighty-seven percent of AR studies involved 'organisational/professional development', or 'educational' settings ; only 13 % were directly 'clinical'. Practitioners were the main participants in 90% of studies. Seventy-two percent of all participant groups were rated 'active' in the research process, yet 70% percent of first (lead) authors were from an academic institution. Patients/carers were generally passive in the research process and absent from authorship. Ninety per cent of studies used two or more methods, predominantly qualitative. Forty-four percent of articles identified external funding sources, relatively high for nursing research. Participatory AR has a strong identity in practice-based research, with a diversity of methods. The focus reflects that of nursing research generally. A high level of participation by practitioners is evident but with little enquiry in authorship. Service user/carer involvement should be given more prominence by researchers.