A total of 492 deaths of children below 5 years of age were registered during a 2-year period of demographic surveillance in a rural population of Ethiopia, where an epidemiologic study base population of 28,780 individuals was established in 1987. Data were collected by lay-reporters using a verbal autopsy method. The under-five cumulative mortality rate was 209 per 1000 children. When sub-divided into infants and children 1-4 years, the respective yearly mortality rates were 101 and 32.3 per 1000. There were considerable variations within the district by Peasants' Associations. Rural Lowlands experienced the highest mortality rates, especially for children 1-4 years. Mortality trends over a 2-year period indicate a significant increase for the child population, but not for infants. Similar trends were observed for boys and girls although the rates for boys were generally higher especially during infancy. More deaths occurred in the months of April, June, and July, and October and November indicating two peak seasons in both years. More deaths occurred in Peasant's Associations that were furthest from the health centre. Major probable causes of death were acute respiratory infections, measles, and diarrhoea. It is concluded that even in rural areas of a developing country it is possible to collect from mothers the much needed and valid fertility and mortality data through epidemiological surveillance by using lay-reporters.