There is continuing controversy over whether the pattern of circadian blood pressure (BP) variation that includes a nocturnal decline in BP and a morning pressor surge has prognostic significance for stroke risk. In this study, we followed the incidence of stroke in 1430 subjects aged > or =40 years in Ohasama, Japan, for an average of 10.4 years. The association between stroke risk and the pattern of circadian BP variation was analyzed with a Cox proportional hazards model after adjustment for possible confounding factors. There was no significant association between total stroke risk and the nocturnal decline in BP (percentage decline from diurnal level) or between total stroke risk and the morning pressor surge. The cerebral infarction risk was significantly higher in subjects with a <10% nocturnal decline in BP as compared with subjects who had a > or =10% nocturnal decline in BP (P=0.04). The morning pressor surge was not associated with a risk of cerebral infarction. On the other hand, an increased risk of cerebral hemorrhage was observed in subjects with a large morning pressor surge (> or =25 mm Hg; P=0.04). Intracerebral hemorrhage was also observed more frequently in extreme dippers (those with a > or =20% nocturnal decline in BP) than dippers (those with a 10% to 19% decline; P=0.02). A disturbed nocturnal decline in BP is associated with cerebral infarction, whereas a large morning pressor surge and a large nocturnal decline in BP, which are analogous to a large diurnal increase in BP, are both associated with cerebral hemorrhage.