In the absence of medical certification of deaths in developing countries, lay reporting and verbal autopsy have emerged as useful alternative methods for collecting data on causes of death. Of these, verbal autopsy offers advantages and is widely used in field studies and child survival programmes. However, because uniform and valid criteria for the diagnosis of common causes of death are lacking, comparison of the results of different studies becomes meaningless. This article proposes such a set of criteria for the cause of death among neonates and for those aged 1-59 months. The criteria are based on the findings of earlier validation studies, a Delphi survey and the experience gained from performing 1000 verbal autopsies in Gadchiroli, India. The emergence of such standardized criteria of causes of death should be of immense value for health planning, monitoring and evaluation purposes and for interregional comparisons.