In this paper we use micro data from a large wage survey in Spain during 2006 to analyze the magnitude of the gender gap in the performance-pay (PP) component of hourly wages. Under the presumption that PP is determined in a more competitive fashion than the other wage components, we argue that there should be less room for gender discrimination in PP. Accordingly, all else equal, the gender PP gap should be low. However, our findings just show the opposite. After controlling for observable characteristics, non-random sorting into PP jobs and segregation into different firms and occupations, the estimated adjusted gap in favour of men remains fairly high (around 30 log points). Further, we document a 'glass ceiling” pattern in the gap throughout the distribution of PP. We examine alternative ways of rationalizing these findings and conjecture that monopsonistic exploitation exerted by employers might be the one that is the most consistent with our evidence.