ABSTRACT Sub-Saharan Africa is thought to be undergoing a health (or epidemiological) and demographic transition, moving from a pattern of disease dominated by infection, perinatal illness and other diseases of poverty, to one dominated by noncommunicable disease, in particular vascular disease. If such a transition is occurring, then the burden of vascular disease including stroke will increase. Stroke is a heterogeneous condition and it is likely that the nature of stroke (pathological types, subtypes, and causes) will change during this transition. However, little is known about the burden and nature of stroke in Sub-Saharan Africa, as it is now. This information is essential to inform health services to appropriately plan and deliver health care for the future, to develop strategies for stroke prevention, and to test the theory of the health and demographic transition. My overall aim was to assess and compare the burden and nature of stroke in rural and urban South Africa, and to establish whether there is any evidence of a health transition. Specifically I aimed to: • review what is known about stroke in Sub-Saharan Africa; • establish the prevalence and nature of prevalent stroke in rural South Africa; • compare the nature of hospital-based stroke in urban and rural stroke patients; • compare the nature of urban hospital-based stroke in different population groups; and • validate two stroke scores in the urban stroke register to enable us to diagnose pathological stroke type in rural stroke patients who do not have access to brain imaging. Methods: The following methods were used to achieve these aims: • I systematically searched the literature for, and critically reviewed, studies of stroke from Sub-Saharan Africa (literature review). • The rural Agincourt Health and Population Unit demographic surveillance site was screened for stroke using two questions during the annual census. Anyone who screened positive for stroke was examined to decide whether they had had a stroke (stroke prevalence study). • The Tintswalo Hospital Stroke Register was established to ascertain and assess rural stroke patients over 20 months (rural hospital-based stroke), and • The Johannesburg Hospital Stroke Register similarly established to assess urban stroke patients over 23 months (urban hospital-based stroke). • The accuracy of the Siriraj and Guy’s Hospital stroke scores was compared to the CT brain scan “gold-standard” in the Johannesburg Hospital Stroke Register. Results: Using these approaches I found that: • Very little is currently known about the burden and nature of stroke in Sub- Saharan Africa. • The prevalence of rural stroke was about half that found in high-income countries, and double that found in Tanzania. However, disabling stroke was at least as prevalent as it is in high-income countries. • Both rural and urban black South Africans are probably in early phases of the health transition, and this is impacting on the nature of stroke, particularly the cause of cardioembolic stroke. • Neither the Siriraj nor Guy’s Hospital stroke score are sufficiently accurate for use in epidemiological studies or clinical management of stroke in Sub- Saharan Africa. Conclusion: There is already a heavy burden of stroke in Sub-Saharan Africa, and there is some evidence of a health transition in the black population. However, it is not possible to accurately assess the burden and nature of stroke without communitybased incidence studies using early brain imaging to distinguish ischaemic stroke from cerebral haemorrhage. Until we have these studies, we will never know the precise burden and nature of stroke, the effect of the health transition, or the optimal approach to preventing a stroke epidemic in our population.