Performance metrics are intended to reinforce accountability in the principal-agent relationship between administrators and elected officials. This form of accountability is often assumed to improve democratic responsiveness and policy outcomes. However, as many scholars have noted, performance measurement systems often result in the intractable problem of administrative gaming, which can dilute accountability in the principal-agent relationship. This qualitative case study generates theory on how and why administrative agents engage in gaming. It develops a typology and discusses the importance of understanding gaming from the perspective of administrators.