The contribution of epidemiological studies of stroke has been fundamental to understanding the occurrence, natural history and the potential for both primary and secondary prevention. Descriptive studies of routinely available data such as death statistics provide some idea of the magnitude of the problem, differences between and within countries, and trends over time. Measuring the incidence of stroke, however, requires special population-based studies which are, unfortunately, expensive and difficult to accomplish and, until recently, provide limited information on the frequency, etiology and natural history of pathologically distinct types of stroke. Analytical studies, in particular the long-established cohort studies, have contributed to an understanding of the risk factors for stroke. Randomized controlled trials on the treatment of raised blood pressure confirmed findings from the observational studies that reducing blood pressure also reduced the relative risk of stroke. Increasing knowledge about stroke subtypes, which may behave in different ways in terms of prognosis and response to therapy, has resulted in further clinical trials tailoring potential therapies according to the precise categorization of subtypes. Few epidemiological studies have been established which can shed light on the reasons for the substantial decline in stroke mortality in recent decades. The available evidence suggests that an improvement in case fatality is the most likely explanation, rather than improvements in incidence. In turn, improvements in case fatality may reflect better management in the acute phase or changes in the natural history of the disease. Epidemiological studies of stroke can contribute to developing more cost-effective strategies for the prevention and control of cerebrovascular disease.