The study compares the cause of death profile in a rural area of South Africa (Agincourt), with that in a rural area of West Africa (Niakhar), and in a developed country with the same life expectancy (France, 1951) in order to determine causes with high and low mortality and priorities for future health interventions. In the two African sites, causes of death were assessed by verbal autopsies, whereas they were derived from regular cause of death registration in France. Age-standardized death rates were used to compare cause-specific mortality in the three studies. Life expectancy in Agincourt was estimated at 66 years, similar to that of France in 1951, and much higher than that of Niakhar. Causes of death with outstandingly high mortality in Agincourt were violent deaths (homicide and suicide), accidents (road traffic accidents and household accidents), certain infectious diseases (HIV/AIDS, tuberculosis, diarrhea and dysentery), certain chronic diseases (cancer of genital organs, liver cirrhosis, gastrointestinal hemorrhage, maternal mortality, epilepsy, acute rheumatic fever, and pneumoconiosis) and malnutrition of young children (kwashiorkor). Causes of death with lower mortality than expected were primarily respiratory diseases (pneumonia, bronchitis, influenza, lung cancer), other cancers, vaccine preventable diseases (measles, whooping cough, tetanus), and marasmus. Verbal autopsies could be used in a rural area of a developing country without formal cause of death registration to identify the most salient health problems of the population, and could be compared with a formal cause of death registration system of a developed country.