Background Sleep duration may be a risk factor for cognitive impairment. Aim To investigate the association between sleep duration and cognitive function in Chilean older adults. Material and Methods We analyzed information from 1,384 participants aged > 60 years participating in the National Health Survey 2009-2010 who were assessed with the Mini Mental State Examination (MMSE) and self-reported their average daily sleep hours. Logistic regression analysis was performed to investigate the association between MMSE and sleep duration. Results Compared to those participants who reported sleeping 7 hours per day, those that reported sleeping < 5 hours had a higher odd for cognitive impairment (Odds ratio (OR): 3.66 [95% confidence intervals (CI: 1.69; 7.95], p < 0.01). Similarly, those who reported sleeping > 8 hours per day also showed a higher odd for cognitive impairment (OR: 2.56 [95% CI: 1.32; 4.95], p < 0.01). This association was even stronger for people who reported more than 10 hours of sleep per day (OR: 4.46 [95% CI: 1.32; 4.95], p < 0.01). Conclusions Long and short sleep duration is associated with cognitive impairment in older adults in Chile independent of major confounding factors.