Background: Perinatal death is a devastating experience for the mother and of concern in clinical practice. Regular perinatal audit may identify suboptimal care related to perinatal deaths and thus appropriate measures for its reduction. The aim of this study was to perform a qualitative perinatal audit of intrapartum and early neonatal deaths and propose means of reducing the perinatal mortality rate (PMR). Methods: From 1st August, 2007 to 31st December, 2007 we conducted an audit of perinatal deaths (n = 133) with birth weight 1500 g or more at Muhimbili National Hospital (MNH). The audit was done by three obstetricians, two external and one internal auditors. Each auditor independently evaluated the cases narratives. Suboptimal factors were identified in the antepartum, intrapartum and early neonatal period and classified into three levels of delay (community, infrastructure and health care). The contribution of each suboptimal factor to adverse perinatal outcome was identified and the case graded according to possible avoidability. Degree of agreement between auditors was assessed by the kappa coefficient. Results: The PMR was 92 per 1000 total births. Suboptimal factors were identified in 80% of audited cases and half of suboptimal factors were found to be the likely cause of adverse perinatal outcome and were preventable. Poor foetal heart monitoring during labour was indirectly associated with over 40% of perinatal death. There was a poor to fair agreement between external and internal auditors. Conclusion: There are significant areas of care that need improvement. Poor monitoring during labour was a major cause of avoidable perinatal mortality. This type of audit was a good starting point for quality assurance at MNH. Regular perinatal audits to identify avoidable causes of perinatal deaths with feed back to the staff may be a useful strategy to reduce perinatal mortality.