Objectives: Obstructive sleep apnea (OSA) affects approximately one third of the population and can reach 90% prevalence in the elderly. There are screening tools to track the disease, however, their performance may differ according to population characteristics. This study aims to determine sensitivity, specificity, predictive value, likelihood ratio and accuracy of the Berlin (BQ) and STOP-Bang (S-Bang) questionnaires and the Epworth Sleepiness Scale (ESS), comparing their performances, using polysomnography (PSG) as a gold standard, in a sample of elderly. Methods: The study was cross-sectional, retrospective, included patients aged 60 or older who underwent PSG type 1, regardless of the BQ, S-Bang and ESS results, during the period of June 1, 2017 to April 30, 2019. OSA diagnosis was by PSG in which the hypopnea apnea index was greater than or equal to 5. Results: Sixty-two patients were evaluated; the prevalence of OSA was 72.58%. The mean age in the sample with OSA was 73.0 sd 8.4 years and without it was 74.7 sd 8.1 years. The sample was predominantly female, 58.1% with OSA. The BQ showed the best results for specificity, predictive value, likelihood ratio and accuracy. S-Bang had the best result for sensitivity and ESS showed the worst results. The BQ odds ratio showed that an individual with a positive BQ has 335% more chance of developing OSA. Conclusion: The QB showed the best performance in the measures for identifying OSA, for a sample of elderly individuals, with a predominance of females and a high prevalence of the disease.