OBJECTIVE: To examine the nature of the relationship between the use of skilled attendance around the time of delivery and maternal and perinatal mortality. METHODS: We analysed health and demographic surveillance system data collected between 1987 and 2005 by the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B) in Matlab, Bangladesh. FINDINGS: The study recorded 59 165 pregnancies, 173 maternal deaths, 1661 stillbirths and 1418 early neonatal deaths in its service area over the study period. During that time, the use of skilled attendance during childbirth increased from 5.2% to 52.6%. More than half (57.8%) of the women who died and one-third (33.7%) of those who experienced a perinatal death (i.e. a stillbirth or early neonatal death) had sought skilled attendance. Maternal mortality was low among women who did not seek skilled care (160 per 100,000 pregnancies) and was nearly 32 times higher (adjusted odds ratio, OR: 31.66; 95% confidence interval, CI: 22.03-45.48) among women who came into contact with comprehensive emergency obstetric care. Over time, the strength of the association between skilled obstetric care and maternal mortality declined as more women sought such care. Perinatal death rates were also higher for those who sought skilled care than for those who did not, although the strength of association was much weaker. CONCLUSION: Given the high maternal mortality ratio and perinatal mortality rate among women who sought obstetric care, more work is needed to ensure that women and their neonates receive timely and effective obstetric care. Reductions in perinatal mortality will require strategies such as early detection and management of health problems during pregnancy.