Background: Heart failure (HF) and cognitive impairment are both common in older adults. However, the association between the two has not been well studied. Methods and Results: We explored the relationship between very probable HF, determined by self-reported symptoms, and cognitive impairment, defined as four or fewer correct on the Six-item Screener, in 14,089 participants of the Reasons for Geographic and Racial Differences in Stroke cohort. We determined the effect of adding demographic, socioeconomic status (SES), health behavior, and comorbidity covariates. In the univariate model, participants with very probable HF were 1.51 (95% confidence interval: 1.15-1.96) times more likely to have cognitive impairment than those without HF. As covariates were added to the model, the relationship between HF and cognitive impairment was attenuated and lost statistical significance after adjustment for depression. Demographic variables, Stroke Belt location (1.28 [1.11-1.48]), SES factors, prior stroke (1.43 [1.18-1.73]), and depression (1.66 [1.38-2.01]) remained significant in the multivariable model. Higher hemoglobin was associated (0.95 [0.9-1.00]) with modestly reduced odds of cognitive impairment. Conclusions: The relationship between cognitive impairment and HF can be accounted for by multiple demographic and SES factors, and by comorbidities, some of which are modifiable. Persons with HF and cognitive impairment should be screened for anemia and depression. © 2008 Elsevier Inc. All rights reserved.