Abstract Background Loneliness is an unpleasant experience linked to multiple adverse health outcomes, but we lack effective interventions to reduce or prevent it. Most interventions are based on a social psychological definition of loneliness (“a perceived discrepancy between actual and desired social relationships, but such interventions are typically unsuccessful. A recent umbrella review of loneliness interventions suggests that they should target specific population subgroups to be effective. This means that we may need more specific definitions of loneliness. In this presentation, I will integrate data from across studies to present an integrated definition of loneliness for older adults. Methods Across two qualitative grounded theory studies, we interviewed two samples of adults aged 60 and over: one living rurally and identifying as socially isolated; the other spousal dementia caregivers whose spouse recently moved to long term care. Data from these studies were integrated and framed in the context of a theoretical synthesis of loneliness we published and recently updated. Results Loneliness was defined in multiple ways, arising from boredom or inactivity, vulnerability, passivity, giving way to predictable behavioural patterns such as distracting tasks, and affective responses such as self-pity. We integrated this with our theoretical synthesis, which frames loneliness as arising from factors on proximate levels (loss of intimate partners, parental relationships, life transitions), as well as more distal influences such as the social network, and culture). Conclusion We argue for cohort-specific definitions of loneliness since these are likeliest to lead to effective interventions. We offer some possible definitions and aspects of loneliness to focus upon in interventions for older adults.