Visual measurements of sleep spindles were carried out in 48 elderly and 20 young normal adults. Computed tomography brain scans and psychometric testing were also performed. Earlier findings of reduced spindle abundance, amplitude and duration in the elderly were confirmed. In addition, we demonstrated a linear increase in spindle density and duration across NREMPs in young adults that was absent in the elderly, indicating that age affects the temporal pattern as well as the quantity of spindles. Contrary to what seemed a highly plausible hypothesis, the amount of waking in the elderly was not inversely correlated with spindle abundance, confirming earlier observations (Feinberg et al. 1967) but in a much larger group. This finding suggests that spindle abundance does not reflect the integrity of the systems that maintain the brain in NREM sleep. We also were unable to show any clear evidence that relative preservation of spindles in the elderly is associated with relative preservation of cognitive skills: psychometric performance and spindle measures were, in most instances, not significantly correlated. However, the test of this hypothesis was limited by the high level of function and the narrow range of impairment of these Ss. One intriguing positive finding was the significant inverse relation between ratings of sulcal atrophy and spindle amplitude. This observation suggests an etiology for the reduced amplitude of the sleep EEG in old age. This change is one of the most striking effects of age on brain electrophysiology.