Relative performance evaluation (RPE) in Chief Executive Officer (CEO) compensation contracts entails the use of peer performance to filter out exogenous shocks and reduce exposure to risk. Theory predicts that high-quality peers can effectively filter out noise from performance measurement, yet prior empirical studies do not examine how differences in peer quality affect the use of RPE in practice. In this study, I propose a model to select peers with the highest capacity to filter out noise and introduce a novel measure of peer quality. Consistent with the theory, I find that firms with high quality peers rely on RPE to a greater extent than firms with few good peers available. I also examine the extent to which peers disclosed in proxy statements overlap with the best peers predicted by my model. I find that the overlap is positively associated with institutional ownership, use of top 5 compensation consultants, and compensation committee competence.