OBJECTIVES: To investigate the effect of cognitive impairment on fatal and nonfatal incident stroke in older adults. DESIGN: A large, national, prospective, population-based study of a representative cohort of older Canadians followed over a 10-year period. SETTING: Secondary analyses were conducted using data from the Canadian Study of Health and Aging, a population-based study of older adults followed prospectively from 1991 to 2001. PARTICIPANTS: Nine thousand four hundred fifty-one adults aged 65 and older who had not previously been diagnosed with stroke at baseline (in 1991). MEASUREMENTS: In addition to known risk factors, the independent contribution of cognitive function (diagnosed in a clinical examination) was examined as a risk for stroke in older adults. RESULTS: Multinomial logistic regression analyses showed that cognitive impairment was associated with twice the odds of fatal incident stroke, controlling for well-established risk factors. CONCLUSION: This study provides further evidence for the need to consider cognitive function in relation to stroke risk in older populations.