Up to one quarter of all strokes are directly attributable to cigarette smoking, which independently increases the relative risk of stroke about three-fold. The risk is dependent upon the amount of cigarettes smoked, is consistent for all subtypes of stroke, and is strongest for subarachnoid haemorrhage and cortical ischaemic stroke caused by arterial atherothromboembolism. The relative risk of stroke is equally high among male and female smokers, and is maximal (compared with non-smokers) n middle age, declining with advancing years. Evidence is also accumulating to implicate pipe and cigar smoking as a risk factor for stroke, and passive exposure to environmental smoke as a risk factor for atherogenesis. The risk of stroke declines considerably and rapidly after stopping smoking, thus supporting a causal relationship, even though it has proved impossible to perform a satisfactory randomised controlled trial. The mechanisms by which smoking causes stroke remain uncertain, but are probably multifactorial and primarily atherogenic.