BACKGROUND: China's success in improving the quality of and access to obstetric care in hospitals offers an opportunity to examine the effect of a large-scale facility-based strategy on neonatal mortality. We aimed to establish this effect by assessing how the institutional strategy of intrapartum care has affected neonatal mortality and its regional inequalities. METHODS: We did a population-based epidemiological study of China's National Maternal and Child Mortality Surveillance System from 1996 to 2008. We used data from 116 surveillance sites in China (37 urban districts and 79 rural counties) to examine neonatal mortality by cause, socioeconomic region, and place of birth, with Poisson regression to calculate relative risks. Rural counties were categorised into types 1-4, with type 4 being the least developed. We report attributable risks and preventable fractions for hospital births versus home births. FINDINGS: Neonatal mortality decreased by 62% between 1996 and 2008. The rate of neonatal mortality was much lower for hospital births than for home births in all regions, with relative risks (RR) ranging from 0·30 (95% CI 0·22-0·40) in type 2 rural counties, to 0·52 (0·33-0·83) in type 4 counties (p<0·0001). The proportion of neonatal deaths prevented by hospital birth ranged from 70% (95% CI 59·7-77·8) to 48% (16·9-67·3). Babies born in urban hospitals had a low rate of neonatal mortality (5·7 per 1000 livebirths); but those born in hospitals in type 4 rural counties were almost four times more likely to die than were children born in urban hospitals (RR 3·80, 2·53-5·72). INTERPRETATION: Other countries can learn from China's substantial progress in reducing neonatal mortality. The major effect of China's facility-based strategy on neonatal mortality is much greater than that reported for community-based interventions. Our findings will provide a great impetus for countries to increase demand for and quality of facility-based intrapartum care. FUNDING: China Medical Board, UNICEF China.