Despite its high prevalence in the aged, hearing loss has been poorly investigated. Audiometry is the gold standard for evaluation of hearing loss, but large-scale use of the procedure involves operational difficulties. Thus, self-report may be an alternative. Aim: To determine if a single global question is valid for use in epidemiologic research. Study design: Systematic review. Material and method: A search of the medical literature from 1990 to 2004 was performed using MEDLINE and LILACS. The references of the articles identified in the electronic search were also reviewed. Study selection and data extraction: The articles that compared the results obtained with self-report to a single global question with those obtained by pure tone audiometry were selected. Data about the prevalence of hearing loss, and sensitivity, specificity and predictive values were extracted. Data synthesis: Ten longitudinal studies were included. A single global question seems to be an acceptable indicator of hearing loss, sensitive and reasonably specific, mainly if the hearing loss is identified as the tone average that includes frequencies up to 2 or 4 kHz, at 40 dBHL level, in the best ear. Conclusion: A single global question shows good performance in identifying older persons with hearing loss and can be recommended for an epidemiologic study if audiometric measurements cannot be performed.