The optimal contacting theory suggests a symmetric relation between CEO pay and firm performance. That is, CEO pay-for-performance sensitivity should remain relatively constant regardless of whether a firm performs good or bad. However, the extant empirical evidence on this issue still reveals lack of consensus. We argue that several factors may increase the heterogeneity of pay-for-performance sensitivity. In this study, we adopt an innovative approach, quantile regression to reexamine how CEO pay-performance sensitivities vary across different levels of CEO pay. We find that conditionally high-CEO equity holding and CEO total pay have a stronger pay-for-performance relation than conditionally low-CEO equity holding and CEO total pay. This suggests that OLS (ordinary least square) may not be an optimal solution to examine this relation. Since the OLS estimate of pay-for-performance sensitivity mainly focuses on the mean relation, and therefore the conclusion based on OLS could be misleading. Additional analyses suggest that large-sized firms demonstrates incremental positive effects of stock return on CEO equity holding. This incremental positive impact of stock return on CEO pay is more pronounced when CEO total pay is used. This could be driven by the fact that pay package for CEOs of small-sized firms comprises a greater portion of cash compensation, attenuating the pay-for-performance sensitivity, and therefore broadening the difference in pay-for-performance between large- and small-sized firms.