Knowledge of the causes of child death is important for health-sector planning since they relate to available interventions. Little is known about causes of child death in Bangladesh from the conventional sources since there is no vital registration system and very few deaths are attended by a qualified physician. To determine the cause structure of child deaths, verbal autopsy interviews were conducted in the Bangladesh Demographic and Health Survey (BDHS) 1993/94 national sample. Verbal autopsy is a method of finding out the causes of death based on an interview with the next of kin or other caregivers. Between BDHS 1993/94 and BDHS 1996/97, 1–4-y-old child mortality in Bangladesh declined by about 27.0%. This impressive decline prompted a verbal autopsy study using the BDHS 1996/97 national sample to determine whether the cause structure had changed. The same verbal autopsy instrument and methods to collect the data and the same computer algorithm to assign causes of death were used in both surveys. Comparison of BDHS 1993/94 and 1996/97 cause-specific mortality rates revealed that deaths due to almost all causes had declined, although significantly so only for acute respiratory infections (ARI), persistent diarrhoea and drowning. Deaths due to neonatal tetanus, acute watery diarrhoea and undernutrition had not decreased at all.

Conclusion: Despite an impressive decline in deaths due to ARI, this condition remains the most important known cause of death in Bangladeshi children. Neonatal tetanus and measles together account for about 10% of deaths in children under 5 y. Further improvements in child survival are possible by improving access to and quality of available child survival interventions.