Objective: To examine the association between breast-feeding initiation time and neonatal mortality in India, where breast-feeding initiation varies widely from region to region.Study Design: Data were collected as part of a community-based, randomized, placebo-controlled trial of the impact of vitamin A supplementation in rural villages of Tamil Nadu, India. Multivariate binomial regression analysis was used to estimate the association between neonatal mortality and breast-feeding initiation time (<12 h, 12 to 24 h, >24 h) among infants surviving a minimum of 48 h.Result: Among 10 464 newborns, 82.1% were first breast-fed before 12 h, 13.8% were breast-fed between 12 and 24 h, and 4.1% were breast-fed after 24 h. After adjusting for birth weight, gestational age and other covariates, late initiators (>24 h) were at B78% higher risk of death (relative risk ¼ 1.78 (95% confidence interval (CI) ¼ 1.03 to 3.10)). There was no difference in mortality risk when comparing babies fed in the first 12 h compared with the second 12 h after birth.Conclusion: Late (>24 h) initiation of breast-feeding is associated with a higher risk of neonatal mortality in Tamil Nadu. Emphasis on breastfeeding promotion programs in low-resource settings of India where early initiation is low could significantly reduce neonatal mortality.