BACKGROUND: Under the Doris Duke Charitable Foundation's African Health Initiative, five Population Health Implementation and Training partnerships were established as long-term health system strengthening projects in five Sub-Saharan Countries. In Zambia, the Centre for Infectious Disease Research in Zambia began to implement the Better Health Outcomes through Mentorship and Assessments (BHOMA) in 2009. This was a combined community and health systems project involving 42 public facilities and their catchment populations. The impact of this intervention is reported elsewhere, but less attention has been paid to evaluation approaches that generate an understanding of the forces shaping the intervention. This paper is focused on understanding the implementation practices of the BHOMA intervention in Zambia. METHODS: Qualitative approaches were employed to understand and explain health systems intervention implementation practices between 2014 and 2016. We purposively sampled six clinics out of the 42 that participated in the BHOMA project within three districts of Lusaka province in Zambia. At the facility-level we targeted health centre in-charges, health workers, and community health workers. In-depth interviews (n = 22), focus group discussions (n = 3) and observations were also collected and synthesised. RESULTS: The major health system challenges addressed by the BHOMA project included poor infrastructure, lack of human resources, poor service delivery, long distances to health centres and inadequate health information systems. In order to implement this in the districts it was necessary to engage with the Ministry of Health and district managers, however, these partners were not actively engaged in intervention design There was great variation in perceptions about the BHOMA interventions. The implementation team considered BHOMA as a 'proof of concept pilot project', running parallel to the public health system, while district health officials from the Ministry of health understood it as a 'long term partner' and were therefore resistant to the short-term nature of the intervention. CONCLUSIONS: The Normalization Process Theory provided a useful framework to understand and explain implementation processes for the BHOMA intervention in Zambia. We clearly demonstrated the applicability of all the four main components of the NPT: coherence (or sense-making); cognitive participation (or engagement); collective action and reflexive monitoring. We demonstrated how complex and dynamic the intervention played out among different actors and how implementation was affected by difference in appreciation and interpretation of the goal of the intervention. Our findings support the growing demand for process evaluations to use theory based approaches to examine how context interact with local interventions to affect outcomes intended or not. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT01942278 . Registered: September 13, 2013 (Retrospectively registered).