Prior research has distinguished general computer self-efficacy (CSE) and application-specific computer self-efficacy, but few studies have investigated the relationship between the two levels of CSE as well as their antecedents and consequences. With the increasing popularity of social networking sites such as Facebook, we are interested in assessing the role of general and specific CSE in predicting users' continuance intention of these sites. An extended technology acceptance model with both cognitive and affective predictors is utilized to capture the decision process of Facebook users' continuance intention. The proposed research model examined antecedents as well as consequences of both general CSE and specific CSE in Facebook. A survey was conducted for data collection. The results show that general CSE and specific CSE are closely related but play different roles in predicting continuance intention in Facebook through cognition and affect, and are affected differently by general computer and Facebook-specific experience.