To study the relation between diurnal blood pressure variations and silent cerebrovascular damage, we performed both 24-hour ambulatory blood pressure monitoring and brain magnetic resonance imaging in 131 elderly asymptomatic hypertensive patients. Silent cerebrovascular damage was identified by the magnetic resonance imaging findings of lacunae (low intensity in T1-weighted images and high intensity in T2-weighted images) and advanced periventricular hyperintense lesions (on T2-weighted images). The frequency of silent cerebrovascular damage in the 100 patients with sustained hypertension was greater than that in the 31 patients with white coat hypertension. We further classified the former group into nondippers (nocturnal reduction of systolic pressure by < 10% of awake systolic pressure; n = 46), dippers (reduction by > or = 10% to < 20%; n = 38), and extreme dippers (reduction by > or = 20%; n = 16). The extent of silent cerebrovascular damage was least severe in the dipper group (P < .05). This J-shaped relation was not found either with the cardiac hypertrophy detected by electrocardiography or with the renal damage assessed by urinary albumin excretion. More than half of the extreme dippers were patients with isolated systolic hypertension, and this prevalence was significantly greater than that in dippers or in nondippers (21% and 30%, respectively). Extreme dippers also had greater variability of pressure (standard deviation of awake systolic pressure) than dippers. Our results indicate that in addition to nondipping, extreme dipping (marked nocturnal fall of blood pressure) should be considered a type of abnormal diurnal blood pressure variation in elderly patients with hypertension who are likely to have advanced silent cerebrovascular damage.