Objective: We aim to evaluate the incidence rate and predictors of clinically significant depressive symptoms (CSDS) over 10 years of follow-up from a population-based cohort study (the Bambui Cohort Study of Aging). Methods: We calculated the predictors of incidence of CSDS over 10 years of follow-up by the Cox proportional regression analysis. Depressive symptoms were evaluated by GHQ-12 and scores of five or higher indicated CSDS. Results: The annualized incidence rate of clinically significant depressive symptoms was 46 per 1000 person-year. In the multivariate analysis, the main predictors of CSDS were cognitive impairment (HR = 1,69 CI95% [1,20 – 2.37], p = 0.002), diabetes (HR = 1.59 CI95% [1.14 – 2.20], p = 0.006), use of 2 to 4 (HR = 1,95 CI95% [1.21 – 3.15], p = 0.006) and of 5 or more medications in the last 90 days (HR = 2.19 CI95% [1.31 – 3.66], p = 0.003) and higher baseline depressive symptoms (HR = 2.12 CI95% [1.61 – 2.78], p < 0.001). Conclusion: These results highlight the importance of higher depressive symptoms, cognitive impairment and endocrine-metabolic disorders to the development of depressive symptoms in older adults. These findings provide a framework for the development of interventions to prevent the emergence of clinically significant depressive symptoms in the elderly.