This study proposes a revised agency theory for the nonprofit sector, distinguishes between the extent of agency and the extent of monitoring, and compares the magnitudes of the two impacts. Using panel data for 1998-2003, the paper tests whether monitoring by principal-stakeholders such as donors, clients, the government, and the board reduces the opportunity for executive misconduct such as extravagant spending on compensation and perquisites. Given the theory, the findings show that two effects influence CEO salaries. First, while nonprofit endowments provide a fiscal cushion in tough financial situations, by offering "organizational slack" they also increase the CEO's opportunity to steal or raise her compensation (i.e., agency effect). Second, donors utilize monitoring mechanisms such as auditing or direct observations, which limit the opportunity for misconduct and reduce executive pay (i.e., monitoring effect). In the final analysis, the monitoring effect is greater than the agency effect, which implies that even if agency problems are present, the monitoring that donors provide offsets them.