The aim of this study was to verify whether social capital is a predictor of all-cause mortality in community-dwelling elderly Brazilians. Participation included 935 surviving elderly from the elderly cohort of the Bambui Project in 2004, who were followed until 2011. The outcome was all-cause mortality and the exposure of interest was social capital, measured in its two components, cognitive (social cohesion and social support) and structural (social participation and neighborhood satisfaction). Sociodemographic variables, health conditions, and smoking were included in the analysis for adjustment purposes. Data analysis was based on the Cox proportional hazards model, providing hazard ratios (HR) and 95% confidence intervals (95%CI). The social participation dimension of social capital’s structural component was the only dimension independently associated with mortality: elderly Brazilians that did not participate in social groups or associations showed a two-fold higher risk of death (HR = 2.28; 95%CI: 1.49-3.49) compared to their peers. The study’s results reveal the need to extend interventions beyond the specific field of health in order to promote longevity, focusing on environmental and social characteristics.