Background: Loneliness in later life is often addressed with befriending interventions, yet evidence for their effectiveness is limited. Meanwhile it is known that loneliness has a deleterious impact on health. The aim of the study is to evaluate whether a befriending service for older adults mitigates the impact of loneliness on health outcomes, and to identify mechanisms through which befriending interventions might impact upon health. Methods: A mixed methods design is used. The quantitative component utilises an AB single-case experimental design, to gather intensive longitudinal data. These data will be analysed using a generalised additive modelling approach. The qualitative component of the study uses semi-structured dyadic interviews, structured and analysed according to the principles of constructivist grounded theory. Findings will then be triangulated according to an existing mixed methods integration protocol. Discussion: This mixed methods design has the potential to inform national and international policy in relation to befriending interventions for older adults. In addition, there is the potential for study results to inform our theoretical understanding of the nature of the relationship between loneliness and health. Trial registration: ClinicalTrials.gov identifier NCT04301167 (10^th March 2020). Protocol version 1.1, 26^th June 2020.