High levels of education have been linked to reduced risk of dementia, whereas magnetic resonance imaging (MRI) white matter hyperintensities (WMH) have been shown to correspond to deficits in executive functioning and psychomotor speed. We studied education, WMH, age, and gender as predictors of better cognitive performance, or cognitive reserve, in the normal elderly. The Repeatable Battery for the Assessment of Neuropsychological Status, supplemented by the Trail Making Test, the Stroop Test, and the California Computerized Assessment Package, were administered to 95 volunteers, aged 75-90 years. Quantitative MRI was used to determine the extent and location of WMH. Using factor analysis, the cognitive measures were reduced to three factors: verbal memory, information-processing speed/executive functioning, and visuospatial skills. When entered into a hierarchical regression, age and gender were the primary predictors of verbal memory, accounting for 34.8% of the variance, with education and WMH adding only 9%. WMH, education, and age contributed independently to predicting speed of information processing/executive functioning, explaining 22.5% of the variance. Only education and age were predictors of visuospatial skills, explaining 14.8% of the variance. These data suggest that cognitive reserve represents a combination of factors that independently determine the threshold for competence within specific cognitive domains.