In a traditional, agricultural society in East Bhutan studies of infant and child mortality were carried out in 1984 and 1991. Mothers were interviewed regarding births during the preceding 5 years and deaths among these children. A significant fall in infant mortality rate (IMR) from 145 to 49 (P < 0.001) was found. In 1991, measles was almost eliminated as a cause of death. Families with 1, 2 and 3 children, respectively, during the period studied, were found to have significantly different IMRs, with higher mortality for higher number of births (P < 0.001 for 1984 and P = 0.002 for 1991). To quantify the effect on mortality from birth intervals both the possibility of reverse causation and confounding factors, such as socio-economic conditions must be considered, but the study suggests that the association is partly causal. The living conditions of the population do not seem to have improved so greatly as to have caused the reduction in IMR. Monthly Mother and Child Health outreach clinics and a network of Village Health Workers have provided the framework necessary for implementation of different health programmes. As a result Primary Health Care has improved during the period between 1984 and 1991, and may have contributed considerably to the reduced mortality.