BACKGROUND: Over one million babies are stillborn or die within the first 28 days of life each year due to preventable causes and poor-quality care in resource-constrained countries. Death audit may be a valuable tool for improving quality of care and decreasing mortality. However, challenges in implementing audit and their subsequent action plans have been reported, with few successfully implemented and sustained. This study aimed to identify factors that affect stillbirth and neonatal death audit at the facility level in the southern region of Malawi. METHODS: Thirty-eight semi-structured interviews and seven focus group discussions with death audit committee members were conducted. Thematic analysis was guided by a conceptual framework applied deductively, combined with inductive line-by-line coding to identify additional emerging themes. RESULTS: The factors that affected audit at individual, facility and national level were related to training, staff motivation, power dynamics and autonomy, audit organisation and data support. We found that factors were linked because they informed each other. Inadequate staff training was caused by a lack of financial allocation at the facility level and donor-driven approaches to training at the national level, with training taking place only with support from funders. Staff motivation was affected by the institutional norms of reliance on monetary incentives during meetings, gazetted at the national level so that audits happened only if such incentives were available. This overshadowed other benefits and non-monetary incentives which were not promoted at the facility level. Inadequate resources to support audit were informed by limited facility-level autonomy and decision-making powers which remained controlled at the national level despite decentralisation. Action plan implementation challenges after audit meetings resulted from inadequate support at the facility level and inadequate audit policy and guidelines at the national level. Poor documentation affected audit processes informed by inadequate supervision and promotion of data usage at both facility and national levels. CONCLUSIONS: Given that the factors that facilitate or inhibit audits are interconnected, implementers, policymakers and managers need to be aware that addressing barriers is likely to require a whole health systems approach targeting all system levels. This will require behavioural and complex intervention approaches.