The aim of this study was to investigate the suitability of the verbal autopsy technique in identification of cause of death during infancy. The study was carried out in 23 randomly selected clusters with 70 live births each, in districts Cuttack and Koraput of Orissa State. Each death occurrence during the reference period (1992-93) was investigated by the field investigator and cross checked by a medical professional using the same proforma. The probable cause of death in each case was analysed separately by the field investigator and then by the professionally qualified medical person. These were then compared using suitable statistical tests. A total of 179 infant deaths and 16 still births were reported by the field investigator in the area as compared to 183 deaths and 12 still births by professionally trained personnel. There was an agreement for 85.79 percent deaths and 75 percent still births. Disagreement was observed for fever, pneumonia, septicemia and meningitis as causes of deaths. It was more in rural area as compared to urban areas. The sensitivity and specificity of the tool were found to be 85.6 percent and 90.4 percent respectively. The technique is found to be cost effective, time saving and reasonably reliable. After intensive training the lay reporters can be used for recording easily recognizable signs and symptoms of infant deaths and thus help in monitoring infant mortality rate and causes of infant death in a vast country like India so as to achieve "health for all" by 2000 AD.