Over the course of 3 years, the authors investigated the relationship between severity of cognitive impairment and mortality in a community sample of 498 elders at high risk for cognitive impairment. Subjects were classified as having no cognitive disorder, mild cognitive impairment, or dementia, based on a validated battery of four neuropsychological tests. Severity of impairment was based on Mini-Mental State Examination scores. Additional data were obtained from subjects' knowledgeable informants and Medicare records. Kaplan-Meier survival estimates and Cox hazard proportion analysis of the sample revealed that presence of cognitive impairment increases mortality in a fashion that parallels the severity of the impairment.