The haemoglobinopathies are the commonest single-gene disorders known, almost certainly because of the protection they provide against malaria, as attested by a number of observations. The geographical distributions of malaria and haemoglobinopathies largely overlap, and microepidemiological surveys confirm the close relationship between them. For two of the commonest disorders, haemoglobin S and alpha(+)-thalassaemia, there is also good clinical evidence for protection against malaria morbidity. However, not all the evidence appears to support this view. In some parts of the world malaria and haemoglobinopathies are not, and never have been, coexistent. It is also difficult to explain why the majority of haemoglobinopathies appear to be recent mutations and are regionally specific. Here we argue that these apparent inconsistencies in the malaria hypothesis are the result of processes such as genetic drift and migration and of demographic changes that have occurred during the past 10,000 years. When these factors are taken into account, selection by malaria remains the force responsible for the prevalence of the haemoglobinopathies.