<jats:p><jats:bold>OBJECTIVES: </jats:bold> To compare dementia risks of elderly black and white subjects and to determine whether differences in education and cognitive test scores contribute to the inconsistency in reported differences between these groups.</jats:p><jats:p><jats:bold>DESIGN: </jats:bold> Longitudinal, 6‐year follow‐up.</jats:p><jats:p><jats:bold>PARTICIPANTS: </jats:bold> Two thousand seven hundred eighty‐six older black and white subjects in the Cardiovascular Health Study.</jats:p><jats:p><jats:bold>MEASUREMENTS: </jats:bold> Age, education (>10 years vs ≤10 years), Modified Mini‐Mental State Examination score (3MS, ≤85 vs >85). Potential confounders were sex, depression, apolipoprotein E4 genotype, vascular disease, and baseline magnetic resonance imaging changes.</jats:p><jats:p><jats:bold>RESULTS: </jats:bold> White subjects with low education and black subjects with high education had twice the risk of dementia of white subjects with high education (95% confidence interval (CI)=1.5–2.4 and 95% CI=1.4–2.7); black subjects with low education had five times the risk of dementia (95% CI=3.4–7.7). Likewise, for subjects with low 3MSE scores, black subjects had 6.7 times the risk of dementia (95% CI=4.7–9.7) and white subjects had 2.7 times the risk of dementia (95% CI=2.2–3.5) as white subjects with high 3MSE scores. Finally, in Cox models, there was no significant black‐white difference in dementia risk after adjustment for all confounders and baseline 3MSE.</jats:p><jats:p><jats:bold>CONCLUSION: </jats:bold> Black race was associated with greater dementia risk even after adjustment for education and other potential confounders. This black‐white difference in dementia risk was markedly attenuated after adjustment for baseline cognitive screening scores. The apparent race effect may reflect gaps in the quality of education or differences in the trajectory of impaired cognitive function experienced by the two groups. Future investigations might take these findings into consideration for the design of studies evaluating black‐white differences in dementia risk.</jats:p>