Objective: This study examined the relationships between social activities, incident cardiovascular disease (CVD), and non-CVD mortality among older adults in the United States. Method: Data from the Health and Retirement Study (2006-2010) were employed. Two measures of social engagement, volunteering and informal helping, along with two measures of social participation, attendance at religious services and social group meetings, were included. Mediation models for health behaviors were estimated. Results: Multinomial logistic regression models demonstrated that volunteering provided the most consistent results in terms of a lower risk of incident CVD and mortality. Furthermore, volunteering at higher time commitments is related to lower CVD incidence and death; informally helping others at a modest time commitment is related to lower risk of death only. Health behaviors mediated the relationships. Social participation was not related to either CVD or mortality. Discussion: Social activity is a modifiable behavior that may be considered a potential health intervention.