We examined the associations between Caesarean section and neonatal mortality in singleton liveborn very-low-birthweight (VLBW) infants (500-1499 g) born during 1984-95 in Washington State, USA, using data from the Washington State birth certificate files. The infants included in this study had no life-threatening congenital malformations and had not been delivered by a repeat Caesarean without a trial of labour (n = 5182). For infants weighing 500-749, 750-999, 1000-1249 and 1250-1499 g, the neonatal mortality rates were 57.8%, 18.6%, 9.7% and 4.7%, respectively, and the Caesarean section rates were 28.4%, 47.8%, 48.0% and 44.6%. The adjusted odds ratios (ORs) for neonatal death associated with Caesarean section were 0.55 [95% confidence interval 0.38, 0.78] for the 500-749 g infants (n = 1059), and 1.15 [0.91, 1.45] for the larger (750-1499 g) infants, after adjustment for birth year, type of hospital, birthweight, presence or absence of labour, breech/malpresentation, and other obstetric indications for Caesarean section (prolapsed cord, placenta praevia, eclampsia, pre-eclampsia and chronic hypertension). However, when the larger (750-1499 g) vertex-presenting (n = 3248) and breech/malpresenting (n = 809) infants were considered separately, the adjusted ORs were 1.42 [1.05, 1.91] and 0.37 [0.23, 0.58] respectively. In contrast, among infants weighing 500-749 g, the ORs were not modified by presentation. The results were similar when we restricted analyses to infants without the above obstetric indications for Caesarean section. Because such an observational study is liable to unmeasurable biases and incomplete reporting of obstetric complications, these OR estimates may be subject to residual confounding. In their present state, these recent population-based data support the view that Caesarean sections do not enhance the neonatal survival of larger (> 750 g) VLBW babies when obstetric complications are absent. The possibility of a protective effect of Caesarean section on the survival of breech/malpresenting infants and infants weighing 500-749 g deserves further studies.