This paper discusses the main changes in infant, child and maternal mortality which have occurred over 1960-1995 in Sub-Saharan Africa and analyses the main factors responsible for the observed shifts in these health trends. To do so, the paper surveys the major mortality models discussed in the literature and appraises their applicability to the Sub-Saharan African situation. Pooled cross section and time series data from 40 African countries, each with a population of over one million people, are then used to estimate explanatory models of infant, under-five and maternal mortality, and of female life expectancy at birth.After controlling for the effects of the time trend, we find that the main determinants of child health status in Africa are real average per capita household income, female literacy rate, nutritional status, safe water supply, immunization coverage and broad accessibility to health services. The health effects of all these factors are statistically significant, though their importance varies according to the population group considered. For instance, income per capita does not appear to be - as often argued - the most important determinant of the health status. Similarly, and somewhat in contradiction with the received theory, female literacy appears to have an impact on infant and child mortality comparable to that of income per capita and access to health care. As expected, immunization has a greater impact on child and maternal mortality than on infant mortality. Calorie supply has a particularly strong impact on under-five mortality but not on infant mortality. Also, maternal mortality falls significantly with the availability of safe water and with increases in vaccinations against neonatal tetanus.The paper also provides evidence that over the period 1960-95 the trend in health status of the African child and female population was not stable, and that an adverse break in trend occurred over the period 1980-95. Specifically, there was a significant upward shift in all mortality trends analysed over this period, while the trend for female life expectancy shifted downward. We interpret this finding as the effect of unfavourable changes (falling incomes per capita, declining health coverage and spread of conflicts and new diseases) which affected most of Africa in the 1980s and early 1990s. Further, we find that there was a change in the structural relationship between per capita income and health status indicators during the period 1980-95. Over that period, a given increase in per capita income was associated with a greater decline in infant, child and maternal mortality than in the previous two decades. We interpret this result as a manifestation of better use of resources in spite of or, rather, because of the increased economic hardships of that period.Our findings reveal the complexity of the evolution of health status in African countries since Independence. The average health outcome for the entire study period (1960-95) is the result of the shifts and breaks in the mortality trend over that time span. It is important therefore to understand the factors responsible for such shifts and breaks in order to effectively use public policy to influence the future course of health status in Africa. Previous economic investigations of mortality trends (see e.g., Musgrove 1987; Anand and Chen 1996) have associated a fall in mortality during periods of economic crisis with the delayed effects of past investments in health care infrastructure. While this might be true also for Africa, it appears that a more efficient use of public health resources (for instance for child immunization) can lead to a fall in mortality even during a recession. Finally, by providing empirical evidence on the relative importance of different policy interventions such as general health service coverage, immunization, provision of water supply, female education and poverty alleviation, the paper can help national and international policy-makers in the selection of the most efficient mix of policy interventions.Apart from providing a framework for analysing the linkages between the evolution of health status and public policy, the paper takes a broader view on mortality determinants. We consider, for instance, the effects of ethnic conflicts on mortality. As one would expect, mortality in Africa is positively correlated with ethnic conflicts; we have however been unable to find statistical significance for the association. The finding is nonetheless important from a policy perspective.The paper ends by providing tentative suggestions on how policy making can help improve health status in Sub-Saharan Africa in the years ahead. The main focus here is on health sector policies, particularly expenditure and pricing policies. The paper argues that these measures should enhance the access to publicly-guaranteed basic health services furnished to all the population by a multiplicity of providers regulated by public authorities. In addition, the paper emphasizes the importance of acting on factors outside the health sector: as conflict prevention and anti-poverty programmes have been found to influence mortality in an important way, policy makers in health ministries need to consider them in formulating overall health care strategies.