Background: Despite increasing use and positive effects of peer support interventions, little is known about how they produce outcomes. Thus, it is essential not only to measure outcomes, but also to identify the mechanisms by which they are generated. Using a realist evaluation approach, we aimed to identify the mechanisms generating outcomes in a Danish peer support intervention for socially vulnerable people with type 2-diabetes (peers). By investigating how the peers interacted in the intervention, we further examined how peers’ individual contextual factors facilitated or hindered the mechanisms in operation.Methods: We used a multi-method case-study design (n=9). Data included semi-structured interviews with four key groups of informants (peer, peer supporter, project manager and a diabetes nurse) for each case (n=25). Furthermore, we collected survey data from peers both before and after participation (n=9). The interview data were analysed using a systematic text condensation, and the Intervention-context-actor-mechanism-outcome framework was used to structure the analysis.Results: We identified two groups of mechanisms that improved diabetes self-management and the use of healthcare services (outcomes): ‘perceived needs and readiness’ and ‘encouragement and energy’. However, the mechanisms only generated the intended outcomes among peers with a stable occupation and financial situation, a relatively good health condition, and sufficient energy (all defined as contextual factors). Independent of these contextual factors, ‘experience of social and emotional support’ was identified as a mechanism within all peers that increased self-care awareness (defined as output). Dependent on whether the contextual factors facilitated or hindered the mechanisms to generate outcomes, we categorised the peers into those who achieved outcomes and those who did not. Conclusions: We identified two groups of mechanisms that improved the peers’ diabetes self-management and use of healthcare services. The mechanisms only generated the intended outcomes if peers’ individual contextual factors facilitated an active interaction with the elements of the intervention. However, independent of these contextual factors, a third group of mechanisms increased self-care awareness among all peers. We highlight the importance of contextual awareness of the target groups in the design and evaluation of peer support interventions for socially vulnerable people with type 2-diabetes.Trial registration: ClinicalTrials.gov, Retrospective Registration (01/20/2021), registration number NCT04722289, https://clinicaltrials.gov/ct2/show/NCT04722289?term=The+Together+on+Diabetes+Intervention+-+a+Realist+Evaluation%2C&draw=2&rank=1