Proactive behavior refers to anticipatory, future- or change-oriented, self-starting behaviors at work. Such behavior is very important for modern organizations because they are facing fast change, fierce competition and great uncertainty. Thus, there has been a surge of research on identifying the predictors of proactivity over the past two decades. Among these, it has been noted that supervisors can play a critical role in promoting employee proactive behavior. Several research gaps, however, are identified after reviewing the extant literature. First, current research has mainly focused on the impact of positive supervisor behaviors such as supervisor support and supervisor's transformational leader behaviors. In reality, supervisors can be abusive or destructive (Einarsen, Aasland, & Skogstad, 2007), and this aspect has not gained adequate attention in the proactivity literature. Additionally, although abusive supervision research has theorized and tested the moderating role of organizational identification in influencing how subordinates respond to supervisory abuse, the results are not consistent. To address these gaps, I draw from the transactional model of stress (Lazarus & Folkman, 1984, 1987), and argue that abusive supervision is negatively related to employee proactive behavior and the interactive effect of abusive supervision and organizational identification on employee proactivity is contingent on individual affective disposition (positive trait affectivity). Thus, there is a three-way interaction of abusive supervision, organizational identification and positive affectivity in influencing employee proactivity. I conducted two multisource studies to test the above hypotheses. In Study 1, I focused on general proactivity, that is, personal initiative. Using data from a sample of 165 dentists and their immediate supervisors from two hospitals, I found that abusive supervision was negatively associated with personal initiative, and when organizational identification was low and positive affectivity was high, the relationship between abusive supervision and personal initiative was the most negative. In Study 2, I used different measures of proactive behaviors (i.e., organizational proactive behavior, supervisory proactive behavior and coworker-directed proactive behavior) to test the robustness and generalizability of the findings in Study 1. Data were collected from 226 employees and their direct supervisors of a large transportation company in China. The results showed that abusive supervision was only negatively associated with supervisory proactive behavior and the three-way interaction of abusive supervision, organizational identification and positive affectivity on three forms of proactive behaviors were all significant. Finally, this dissertation summarizes the main findings, discusses the theoretical and practical implications, and indicates limitations and future research directions. Keywords: Proactive behavior, Abusive supervision, Organizational identification, Positive affectivity