Background : During the COVID-19 pandemic ‘social distancing’ has highlighted the need to minimise loneliness and isolation among older adults (aged 50+). We wanted to know what remotely delivered befriending, social support and low intensity psychosocial interventions may help to alleviate social isolation and loneliness and how they work. Methods : We followed a systematic ‘review of reviews’ approach. Searches of 11 databases from the fields of health, social care, psychology and social science were undertaken during April 2020. Reviews meeting our PICOS criteria were included if they focussed on the evaluation of remote interventions to reduce levels of social isolation or loneliness in adults aged 50+ and were critically appraised using AMSTAR2. Narrative synthesis was used at a review and study level to develop a typology of intervention types and their effectiveness. Intervention Component Analysis (ICA) and Qualitative Comparative Analysis (QCA) were used at a study level to explore the characteristics of successful interventions. Results : We synthesised evidence from five systematic reviews and 18 primary studies. Remote befriending, social support and low intensity psychosocial interventions took the form of: (i) supported video-communication; (ii) online discussion groups and forums; (iii) telephone befriending; (iv) social networking sites; and (v) multi-tool interventions. The majority of studies utilised the first two approaches, and were generally regarded positively by older adults, although with mixed evidence around effectiveness. Focussing on processes and mechanisms, using ICA and QCA, we found that the interventions that were most successful in improving social support: (i) enabled participants to speak freely and to form close relationships; (ii) ensured participants have shared experiences/characteristics; (iii) included some form of pastoral guidance. Conclusions : The findings highlight a set of intervention processes that should be incorporated into interventions, although they do not lead us to recommend specific modes of support, due to the heterogeneity of interventions.