Background: In 2005 Bennett proposed a conceptual model of cognitive deficits (CD) in heart failure (HF). The purpose of this study was to test the model and to identify if results differ by gender. Methods: Data from 280 HF patients were analyzed using generalized linear modeling to predict CD measured as the number of neuropsychological tests on which participants scored in the abnormal range. Five tests (Trail-Making Tests A and B, Digit Symbol-Substitution, Probed-Recall Memory, Letter Number Sequencing) were used; the number on which subjects scored below their age-based norm was used as the measure of CD (0-5 range). Predictors entered into the model were age, sex, education, Charlson Comorbidity score, hypertension (yes/no), depression measured with the PHQ-9, HF type (systolic/mixed or diastolic), HF etiology (ischemic/nonischemic), ejection fraction, NYHA functional class, duration since diagnosis, and number of medications taken daily. Only oxygen saturation proposed in the original model was unavailable. Results: Overall, the model was adequate to describe CD but several of the variables were poor predictors so the model was rerun separately in men (n=180, mean age 63±12.3, 33% black) and women (n=100, mean age 60±12.7, 36% black). In men the model fit well (p=0.005) but not in women (p=0.969). In men the best indicators of CD were older age, less education, and poor functional class; these variables explained 36.3% of the variance in CD. In women, a new model was proposed that focused on behavior (smoking, drinking, exercise, medication adherence, taking vitamins, body weight, sleep hygiene), symptoms (NYHA functional class, sleep quality, fatigue, daytime sleepiness, depression), and social support. Demographic characteristics of age and education were included, as these are known to predict CD. In women the best predictors of CD were older age, lower medication adherence, worse NYHA functional class, and less social support. These 4 variables explained 55.9% of the variance in CD in women. When the new model was tested in men, only NYHA class was associated with CD, explaining 1.5% of the variance. Conclusion: Different factors are associated with cognitive decline in men and women. Different interventions targeting CD may be needed.