The mortality rate of stroke has decreased dramatically over the past 30 years in Japan, but the prevalence of cerebral infarction has increased, mainly because of the higher incidence of nonfatal cerebral infarction in the elderly. Significant advances in the development of antihypertensive drugs over the last 30 years have contributed to the decrease in stroke mortality, but antihypertensive treatment appears to relate to an increase in the prevalence of cerebral infarction in the elderly. We investigated the risks of low BP levels determined by ambulatory BP monitoring and home BP measurements to clarify the previously reported J-shaped relationships between the incidence of ischemic cardiovascular disease and BP levels after treatment. A long-term prospective study of all causes of mortality and cerebrovascular and cardiovascular mortality has been conducted in Ohasama, Japan, since 1987. Cox proportional hazard model after adjusting for age and sex demonstrated that during a 5-year follow-up period involving 893 of the subjects aged 50 years and over in this cohort, those with the lowest quintile of ambulatory BP levels exhibited a significantly high hazard ratio of cardiovascular as well as all causes of mortality. During the same follow-up period in 1,226 subjects aged 50 years and over, those with the lowest and highest quintiles of home BP levels demonstrated a significantly high hazard ratio of cardiovascular as well as all causes of mortality (i.e., J-shaped relationship). The amplitude of the fall in nocturnal BP obtained by ambulatory BP monitoring correlated positively with the severity of silent cerebrovascular lesions in elderly women, but not in elderly men. These results suggest the significantly high risk associated with low BP levels, which can be determined only by ambulatory and home BP measurements, but not by casual BP measurements.