OBJECTIVE: The authors compared amounts of white matter hyperintensity in late- and early-onset depressed patients and never-depressed older subjects, compared neuropsychological function in these groups, and investigated the association between white matter hyperintensities and cognitive function in depression. METHOD: Sixty currently depressed patients whose first depression occurred after age 50 years, 35 depressed patients over age 50 whose first depression occurred before age 35, and 165 nonpsychiatrically ill subjects over age 50 underwent magnetic resonance imaging (MRI) and neuropsychological evaluation. Areas of white matter hyperintensity were measured from MRI images. RESULTS: The late-onset patients had more white matter hyperintensity than either of the other groups. Compared to the nondepressed subjects, the patients had significantly lower scores in the cognitive domains of nonverbal intelligence, nonverbal memory, constructional ability, executive ability, and information processing speed. The cognitive abnormalities were mostly confined to the late-onset patients, and the presence of a large amount of white matter hyperintensity was associated with significantly poorer executive skills. However, most of the scores were not in the significantly impaired range. CONCLUSIONS: Large amounts of white matter hyperintensity are more frequent in patients with late-onsetdepression than in elderly subjects with early- onset or no depression. Both late- and early-onset elderly depressed patients show mild decrements in some "right hemisphere" cognitive skills; the late-onset subjects also show deterioration in information processing speed and executive functions. Patients with large amounts of white matter hyperintensity have significantly poorer executive function.