A systematic review of the research literature using Medline, Embase, Psyclit/Psycinfo and the Cochrane Library files 1980 through 2001, identified only eight publications based on well-designed studies involving randomised controlled trials (RCTs)--testing the effects of modification of provider-patient interaction and provider consulting style on patient diabetes self-care and diabetes outcomes, in general practice or hospital outpatient settings. Review of these publications leads to the tentative conclusion that focusing on patient behaviour--directly enhancing patient participation i.e. by assistant-guided patient preparation for visits to doctors, empowering group education, group consultations, or automated telephone management--is more effective than focusing on provider behaviour to change their consulting style into a more patient-centred one. The latter proves hard to sustain, needs intensive support, and is not very effective in improving patient self-care and health outcomes when executed alone. Patient behaviour focused interventions show good efficacy and efficiency, and improve patient self-care and diabetes outcomes. More well-designed intervention studies focusing on enhancing patient participation in primary and hospital outpatient diabetes care are needed.