Background: The 12 item-General Health Questionnaire (GHQ-12) is the shortest version of GHQ. The questionnaire works as well as the longer instruments and is used more frequently in epidemiological surveys. It is an effective instrument for aging studies because of its brevity and absence of somatic symptoms. Previous studies of Principal Component Analysis (PCA) were carried out in adult populations only, demonstrating two- and three-factor solutions. Methods: This study involved 392 participants over 75 years old in a population-based survey in Bambui, Brazil. Botn PCA and Confirmatory Factor Analysis (CFA) were performed. In CFA, several factor structures were compared using different goodness-of-fit indices. (The association between factor scores of the model with the best fit and participant characteristics were analyzed by linear regression). Results: PCA suggested a three factor model. Graetz's three factor solution (Anxiety Depression, Social Dysfunction and Loss of Confidence) was the best model in CFA according to goodness-of-fit indices. Activity limitation and poor self-reported global health were associated with the anxiety/depression and social dysfunction factors. Cognitive impairment and female gender were associated with social dysfunction. Loss of confidence was not associated with these or other relevant variables. Conclusion: The three-factor solution proposed by Graetz seems to be the best fit also for people over 75 years old. Further studies are needed to understand the conceptual and practical relevance of these underlying factors, particularly loss of confidence. in the elderly population