We prospectively tested in the combined original and offspring Framingham cohorts the hypothesis that the increase in cardiovascular disease (CVD) incidence at low diastolic blood pressure (BP) is largely confined to subjects with increased systolic BP and hence an increased pulse pressure. The 10-year risk of 951 nonfatal CVD events and 204 CVD deaths was estimated at diastolic pressures of <80, 80 to 90, and > or =90 mm Hg, according to concomitant systolic BP. An increasing tendency for a J-curve relation of CVD incidence to diastolic BP was observed with successive increments in accompanying systolic BP. In both genders, a statistically significant excess of CVD events was observed at a diastolic BP of <80 mm Hg only when accompanied by a systolic BP of >140 mm Hg that persisted after adjustment for age and associated CVD risk factors. Patients with this condition of isolated systolic hypertension have been shown to benefit from antihypertensive treatment.