To date, the suitability of affiliative behaviours for inclusion in a horse welfare assessment protocol has not been thoroughly studied. This is surprising, given the growing focus on demonstrating the presence of positive welfare rather than just the absence of poor welfare. This shift is driven by the recognition that simply avoiding negative experiences does not necessarily equate to a positive welfare state, and that there is more to good welfare than just the absence of disease, injury or stress. Thus, we critically reviewed the current scientific literature to propose quantitatively assessed animal-based indicators that indicate positive welfare and could therefore be incorporated in a welfare assessment protocol for horses. Amongst the candidate indicators that have previously been associated with positive welfare in horses are socio-positive interactions between horses, such as affiliative activities consisting of social proximity, social play and social grooming. These activities must meet the scientific quality criteria regarding validity, reliability and feasibility to enable an objective assessment of positive animal welfare. However, the current review showed that social play and social grooming amongst adult horses under husbandry conditions are not sufficiently valid indicators. These two affiliative behaviours would meet the requirement for validity only in combination with a differential diagnosis considering the absence of, for example, the stress caused by space restriction. Furthermore, social play and social grooming amongst adult horses may not fulfil the requirement for feasibility, because these behaviours occur rarely and unpredictably throughout the day, hindering their assessment within a limited observation time. As a result of this literature review, we suggest that social proximity may be the only suitable affiliative behaviour indicating positive welfare in adult horses. It best reflects the social bond with preferred partners - a key element for experiencing positive welfare. This affiliative behaviour has been investigated thoroughly in various sport and leisure horse housing systems regarding all three scientific quality criteria. It satisfies the validity criterion because to be counted as a socio-positive interaction, social proximity must be achieved voluntarily whereby the term ‘voluntary social proximity’ may best fit this requirement. Notably, it also fulfils the reliability and feasibility criteria in terms of good inter-observer agreement given a standardised assessment method including parameters such as distance between horses, duration and context. Thus, social proximity seems to be a promising indicator of positive welfare in horses.