PURPOSE: To examine relationships between brain white matter hyperintensities depicted at magnetic resonance (MR) imaging and performance on neuropsychologic tests in community-dwelling elderly adults. MATERIALS AND METHODS: The 1921 Aberdeen Birth Cohort is a subsample of survivors of the Scottish Mental Survey of 1932 whose mental ability was tested at 11 years of age. Ninety-five of these subjects agreed to undergo brain MR imaging, an examination of general health, and a neuropsychologic evaluation. White matter hyperintensities detected at T2-weighted MR imaging were rated by using a semi-quantitative method yielding two continuous variables: white matter lesions and periventricular lesions. Cognitive ability, including crystallized and fluid intelligence domains, was assessed with standard neuropsychologic tests. RESULTS: Rating scores of white matter lesions were normally distributed (on a devised scale) with means of 1.14 for white matter lesions and 1.28 for periventricular lesions. Intra- and interobserver reliability coefficients for scores were high, generally above 0.7. There were significant correlations of medium effect size between the T2-weighted MR imaging-depicted white matter lesions and performance on tests of fluid-type intelligence. No significant correlation was demonstrated between white matter lesion ratings and tests of crystallized intelligence. CONCLUSION: Lower fluid-type (“prevailing”) intelligence test scores were associated with increased severity of white matter lesion ratings but not crystallized type (“remorbid”) intelligence test scores. This indicates that MR imaging-depicted white matter lesions are of clinical importance.