Treatment of stroke patients in specialised stroke units has become more frequent, but the longterm effect of this treatment has not been determined. In this prospective, community-based study of 1241 unselected acute stroke patients we compared outcome between patients geographically randomised to treatment in a stroke unit or in a general neurological/medical ward, from the time of acute admission to the end of rehabilitation. Baseline characteristics were comparable between the two treatment groups regarding age, sex, marital status, pre-stroke residence, and stroke severity. Patients treated in the stroke unit had higher comorbidity with regard to hypertension and diabetes. Multivariate linear and logistic regression analyses were applied to estimate the independent influence of stroke unit treatment on outcome. Stroke unit treatment significantly reduced not only initial mortality, but also mortality within five years from stroke onset. The relative risk of dying within the first five years from stroke was reduced by 40%, p < 0.01. Treatment and rehabilitation of unselected stroke patients in a stroke unit reduces initial mortality, discharge rate to nursing home, reduces cost of treatment, and improves longterm survival up to five years after stroke.