AIM: To compare causes and rates of mortality among infants admitted to 10 Australian neonatal intensive care units (NICUs) between 1995 and 2006. METHODS: De-identified perinatal data from the Neonatal Intensive Care Units' (NICUS) Data Collection for 24 131 infants were examined for causes and rates of death. The study period was divided into two epochs: I (1995-2000, n = 11 185 infants) and II (2001-2006, n = 12 946 infants). RESULTS: A total of 2224 (9.2%) infants died in hospital. Mortality decreased from 10.3% (1152/11 185) in epoch I to 8.3% (1072/12 946) in epoch II (p < 0.001) due to improved survival in term infants. Extreme prematurity also decreased as a primary cause of death (118 (10.2%) vs 76 (7.1%), p = 0.008). No infant >42-week gestation was admitted in epoch II. Congenital abnormalities were the most common cause of death (>20%) in both epochs, mostly in term rather than preterm infants (40.7% vs 13.9%, p < 0.001). Age of death was unchanged between the two epochs (median 4, 1st, 3rd quartiles: 1,16 days). CONCLUSION: Mortality rates have continued to decrease but improvement is predominantly due to improved survival of term infants and prevention of postdate deliveries. Congenital abnormalities continue to be the most common cause of death.