The study compares the cause of death profile in a rural area of South Afri
ca (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 t
o determine causes with high and low mortality and priorities for future he
alth 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 hig
her than that of Niakhar. Causes of death with outstandingly high mortality
in Agincourt were violent deaths (homicide and suicide), accidents (road t
raffic accidents and household accidents), certain infectious diseases (HIV
/AIDS, tuberculosis, diarrhea and dysentery), certain chronic diseases (can
cer of genital organs, liver cirrhosis, gastrointestinal hemorrhage, matern
al mortality, epilepsy, acute rheumatic fever, and pneumoconiosis) and maln
utrition of young children (kwashiorkor). Causes of death with lower mortal
ity than expected were primarily respiratory diseases (pneumonia, bronchiti
s, influenza, lung cancer), other cancers, vaccine preventable diseases (me
asles, whooping cough, tetanus), and marasmus. Verbal autopsies could be us
ed in a rural area of a developing country without formal cause of death re
gistration to identify the most salient health problems of the population,
and could he compared with a formal cause of death registration system of a
developed country.