Background: Failure to keep people living with HIV engaged in life-long care and treatment has serious implications for individual and population-level health. Nested within a four-province study of HIV care and treatment outcomes, we explored the dynamic role of social and service-related factors influencing retention in HIV care in Zambia. Methods: From a stratified random sample of 31 facilities, eight clinics were selected, one urban and one rural from each province. Across these sites we conducted a total of 69 in-depth interviews, including with patients (including pregnant women) engaged in-care (n = 28), disengaged from care (n = 15), engaged facility transferee (n = 12), and friends/family of deceased patients (n = 14). At the same sites we conducted 24 focus group discussions with a total of 192 lay and professional healthcare workers (HCWs). Two-day observations in each of the eight facilities helped triangulate data on operational context, provider relations and patient-provider interactions. We ordered and analysed data using an adapted version of Ewart's Social Action Theory. Results: Three overarching findings emerged. First, the experience of living with HIV and engaging in HIV care in Zambia is a social, not individual experience, influenced by social and gendered norms and life goals including financial stability, raising family and living stigma-free. Second, patients and their networks act collectively to negotiate and navigate HIV care. Anticipated responses from social network influenced patients' willingness to engage in care, while emotional and material support from those networks influenced individuals' capacity to remain in HIV care. Lastly, health system factors were most influential where they facilitated or undermined peoples' collective approach to health service use. Participants living with HIV reported facilitation of both their initial and continued engagement in care where services involved social networks, such as during couples testing and community outreach. Conversely, service features that were poorly aligned with respondents' social reality (e.g. workplace obligations) hindered long-term engagement. Conclusions: This study moves beyond listing barriers or socio-ecological groupings, to explain how social and health systems interact to produce HIV care outcomes. Our findings challenge the implicit assumption of individual agency underpinning many retention studies to highlight the social nature of illness and healthcare utilization for HIV in Zambia. This understanding of collective action for accessing and remaining in HIV care should underpin future efforts to revise and reform HIV and potentially other chronic service models and systems.