Previous research found an association between sleep disturbances and cognitive deficits on the one hand, and between increased cortisol levels and poor cognitive performance on the other hand. We hypothesized that cortisol may, at least partially, mediate the link between sleep disturbances and cognitive impairment (CI). We analyzed data from 440 nondemented subjects aged ≥65 years (72.4 ± 4.5 years old, 55.7% women) participating at the population-based CoLaus/PsyCoLaus study, who underwent cognitive evaluation, complete polysomnography and cortisol measures during the day. Subjects with CI (N = 207, 47.05% of the sample) had lower sleep efficiency, less deep sleep (stage N3) and rapid eye movement sleep, and higher apnea/hypopnea index and oxygen desaturation index. After adjustment for possible confounders, oxygen desaturation index (≥4% and ≥6% per hour of sleep) were significantly associated with impaired cognitive performance. The results of Sobel's test for mediation using the regressions between the sleep-related variables and cortisol values, and between the cortisol and the Clinical Dementia Rating score were not significant (all p > 0.05). Our data suggest that sleep-disordered breathing is associated with CI, but that this association is not mediated by increased diurnal cortisol levels.