Nondemented older subjects with cognitive impairment are at higher risk of death than are mentally intact persons. It is unknown whether the effect of cognitive impairment on mortality is independent or mediated by poorer health or other factors. To address this issue, the authors used a prospective population-based cohort study with 8 years of follow-up started in 1987 in the urban district of Kungsholmen, Stockholm, Sweden. Subjects were all 75- to 95-year-old subjects nondemented at baseline (n = 1,435). Data on cognition, health (mental and subjective health, physical diseases and comorbidity, nutritional status, and disability), and apolipoprotein E genotype were collected at baseline as was time of death at follow-up. Cognitive impairment was defined on the basis of age- and education-specific cutoffs on the Mini-Mental State Examination. When compared with cognitively unimpaired subjects (n = 1,210), those cognitively impaired (n = 225) had poorer health conditions, were more frequently apolipoprotein E epsilon4 carriers, and had greater mortality risk (hazard ratio = 1.7; 95% confidence interval: 1.5, 2.1). Neither health nor apolipoprotein E genotype accounted for the association of cognitive impairment with mortality (adjusted hazard ratio = 1.6; 95% confidence interval: 1.3, 1.9). The functional and behavioral consequences associated with the progression of cognitive impairment may mediate the increased mortality risk.