This paper reviews the methodological issues in the epidemiological study of the outcome from stroke. Data are presented from an unselected series of patients in whom the underlying stroke pathology is clearly defined. Although the natural history varies among the different pathological subtypes of stroke, simple clinical baseline measures of the severity of the neurological deficit (incontinence, loss of consciousness and severity of paresis) and premorbid level of disability and social functioning independently predict disability-free survival by 1 year, and may help direct management and research.