Systematic, computerized search in Medline, the Cochrane Library, Eric, PsychINFO and Embase files, 1980-2003, selecting descriptions of prospective intervention trials with good methodological design, testing effects of social support interventions on health outcomes in primary and outpatient care for type 2 diabetes. Six controlled trials were reviewed. They defined, modified, and measured social support in various ways, and scored outcomes with varying measures. Gender differences and the right amount of support seem important. Promising new forms of social support: group consultations (better HbA1c and lifestyle), Internet or telephone-based peer support (improved perceived support, increased physical activity, respectively), and social support groups (improved knowledge and psychosocial functioning). No improved diabetes control by classic forms of support, e.g. from spouse (but weight loss in women) and family and friends (no differences). It is tentatively concluded that this review supports the hypothesis that specific social support interventions affect patient self-care and diabetes outcomes. New forms of social support may be discussed and incorporated in the work of diabetes teams, and offered to patients as new possibilities to help them adjust to a life with (type 2) diabetes and make information-based decisions. Only in the group consultations study, diabetes control was protected. More well-designed research testing the effects of specific social support interventions on patient self-care, lifestyle adaptations, and outcomes of diabetes care, is warranted.