Employees are directly involved in work tasks and processes which are necessary to accomplish unit or organizational goals, and accordingly, they may become aware of key mistakes, slips, and failures that are unbeknownst to the leader or supervisor responsible for the work unit or organization. Given that errors or deviations in work tasks or processes can have far-reaching effects within the organization, it may be essential for employees to share bad news with their leader or supervisor so that steps can be taken to address the issue or ameliorate negative consequences. However, although employees' sharing of bad news may be important to the organization and should be encouraged, supervisors may respond to the messenger in ways that discourage the behavior. Unfortunately, we lack an explanation of why and under what conditions supervisors respond positively or negatively to employees who share bad news. Thus, the purpose of this dissertation is to address this gap in our understanding. I draw from social exchange theory and the transactional theory of stress to develop a conceptual model of sharing bad news. I suggest that sharing bad news can be cast as a transaction between employees and supervisors that is mediated by supervisors’ appraisals of employees’ sharing the message. The quality of the relationship between an employee and supervisor, or leader-member exchange (LMX), is strengthened when supervisors appraise the sharing of bad news as challenging, or potentially rewarding; however, LMX is weakened when supervisors appraise the sharing of bad news as hindering, or potential harmful. In turn, LMX influences supervisor responses to the sharing of bad news in the form of evaluations of the employee’s effectiveness. In addition to these main effects, I also consider how aspects of the message delivery, such as the timeliness with which messages are conveyed and extent to which employees incorporate solutions when they share bad news, can influence supervisor appraisals of sharing bad news. Finally, I suggest that the extent to which the messenger is responsible for the bad news moderates the relationships between appraisals of sharing bad news and LMX. I test this model in three studies.