OBJECTIVE To assess risk factors for neonatal mortality in a tertiary level neonatal unit. METHODS Case control analysis of routine neonatal data for 1998. METHODS Harare Central Hospital Neonatal Unit. METHODS All neonates delivered at Harare Maternity Hospital and admitted to the neonatal unit for care between January and December 1998. METHODS Neonatal mortality in hospital. RESULTS A total of 5,305 neonatal admissions were studied of which 19.3% died in hospital. The median age at death was two days (Q1 = 1, Q3 = 3) and the median age at hospital discharge was 3 days (Q1 = 1, Q3 = 6). Risk factors for mortality were un-booked mother odds ratio (OR) 2.36 (95% CI = 1.98 to 2.81), breech delivery OR: 1.76 (95% CI = 1.39 to 2.22), low birth weight OR: 4.67 (95% CI = 3.92 to 5.57), prematurity OR: 2.36 (95% CI = 2.09 to 2.66), congenital malformations OR: 2.80 (95% CI = 1.72 to 4.53) and birth asphyxia OR: 1.79 (95% CI = 1.51-2.12). Being admitted for respiratory distress was associated with better survival OR: 0.22 (95% CI = 0.17 to 0.28). Having a Caesarian section was also protective OR: 0.60 (95% CI = 0.47 to 0.76). Mother's age, parity, time of delivery and sex were not significantly associated with mortality odds ratios (95% CI) of 1.07 (0.86 to 1.34), 0.94 (0.78 to 1.13), 1.10 (0.93 to 1.30) and 0.89 (0.78 to 1.03) respectively. On regression analysis birth weight greater than 2,500 g, being un booked and breech delivery were predictive of mortality with OR (95% CI) of 0.99 (0.99 to 0.99), 1.31 (1.12 to 1.61) and 1.15 (1.04 to 1.28) respectively. CONCLUSIONS Low birth weight is the highest risk factor for mortality in this tertiary level hospital. Strategies targeted at low birth weight infants such as antenatal corticosteroid use, improved intrapartum care, appropriate antibiotic use, improved efficiency and access to neonatal intensive care will have the most impact on neonatal mortality.