PURPOSE: The objective of this study was to investigate whether the potentially inappropriate medication (PIM) use is a predictor for mortality in a community-based population of older adults. METHODS: A cohort study was performed between January 1, 1997, and December 31, 2011, based on data from a representative sample of the population aged 60 years or older living in Bambuí, Brazil. Univariate and multivariate analyses of the association between the variable of interest (use of PIM or pharmacological groups of PIM) and mortality were based on the extended Cox model for proportional hazards. The extended model was adopted to include the measurement of the exposure of interest throughout the follow-up period and not only at baseline. Adjustment variables included sociodemographic characteristics, health status, and use of health care services and medications. A level of significance of 5% was adopted for all analyses. RESULTS: The prevalence of PIM use was 56.0% (95%CI 53.4-58.6). After multiple adjustments, the risk of death among users of at least one PIM was 44% higher (HR = 1.44; 95%CI 1.21 to 1.71) than among those who did not use any PIM. Among the PIM groups analyzed, antipsychotics were the most strongly associated with mortality (HR = 2.33; 95%CI 1.72 to 3.17). CONCLUSIONS: The study revealed a high prevalence of PIM use among the community-based aged population. It also identified PIM use as a predictor for mortality in this group of individuals. This indicates the need for selection of safer drug therapy alternatives in this patient group.