OBJECTIVES To determine the prospective relationship between blood pressure (BP) and cognitive function across a wide age range. DESIGN Prospective population-based cohort study. SETTING The Rotterdam Study and the Leiden 85-plus Study. PARTICIPANTS Three thousand seventy-eight men and women, initial age 55 to 84 from the Rotterdam Study and 276 men and women, initial age 85, from the Leiden 85-plus Study. MEASUREMENTS Systolic BP (SBP) and diastolic BP (DBP) were measured at baseline, cognitive function was assessed at the end of follow-up using a dedicated neuropsychological test battery. The association between baseline BP levels and cognitive function later in life was assessed in 10-year age groups in the Rotterdam Study and in 85-year-olds of the Leiden 85-plus Study. RESULTS In the youngest participants (< 65), SBP and DBP were not associated with cognitive function 11 years later. For persons aged 65 to 74, higher baseline SBP and DBP were related to worse cognitive function 11 years later. In contrast, in older age (>= 75), higher SBP and DBP seemed to be related to better cognitive function at the end of follow-up. This effect appeared strongest in the highest age group (aged 85). CONCLUSION High BP was associated with greater risk of cognitive impairment in persons younger than 75 but with better cognitive function in older persons. Age-specific guidelines for BP management are needed, because the current directive that "lower is better" may not apply to BP levels in the very old.