Online Health Communities (OHCs) have become an important source of sharing and receiving information and support for people with health-related concerns. These communities provide important benefits to users including enhanced medical knowledge, emotional comfort, personal empowerment and the ability to create offline social connections. High levels of user engagement are beneficial to both users and the OHC, so it is important to understand what motivate users’ participation, encourage them to contribute and influence their churning behaviors. This thesis covers why, when, and how users are actively engaged within an OHC. It is based on descriptive and predictive analytics of OHC users’ online interactions with text mining techniques. I built explanatory models to reveal how users’ motivations and roles evolve over time, the types of social support activities that encourage users’ continuous participation, and the forms of social capital that drive users’ continued contributions to the community. In addition, I developed predictive models to help an OHC forecast whether and when a user will churn. The findings of this study have implications for managing and sustaining successful OHCs, and can provide OHC managers with suggestions on how to motivate user contributions and retain users through interventions.