The Securities and Exchange Commission’s 2006 Executive Compensation Disclosure Rules require firms to disclose how executive pay is determined by benchmarking total compensation at the competitive labor market level (compensation benchmarking) and by benchmarking performance targets in relative performance evaluation (performance benchmarking). Prior studies examining the selection of peer firms typically focus on one or the other benchmark. Using Incentive Lab’s detailed data on proxy statements from 2006 through 2015, we find that approximately 57% of the peers are used simultaneously for both compensation and performance benchmarking, a pattern largely ignored in prior literature. We label these peers as “dual-role peers” and show that firms can indeed succeed in selecting such peers in order to achieve high pay and yet low expected performance. Moreover, we find that the extent of such discretionary peer selection is positively associated with realized excess CEO pay, and negatively associated with ex-post stock performance in the subsequent year. Additional evidence shows that the power of CEOs to intervene the boards’ compensation decisions exacerbates the opportunistic peer selection. Our study provides new evidence on managerial self-serving behavior in compensation practices and highlights the importance of considering dual-role peers in compensation research.