This article seeks to determine whether discriminators perform differently than nondiscriminators, whether that performance difference serves to inhibit discrimination, and, if so, how the process works. It investigates an area with abundant data on both the racial mix of participants and firm performance-major league baseball. A sizeable and statistically significant relationship between winning and the presence of black players in the starting Lineup is found to hold for the early years of the 1950s and to decline, as expected, as the average number of black ballplayers on major league rosters increased. However, a puzzle emerges in the fact that a significantly slower rate of integration was pursued in the American League (AL) than in the National League (NL) despite there being nearly identical associations between black players and winning in each. The most likely solution to that puzzle is that fan preferences differed-each black player on an AL team appears to have reduced attendance by significantly more than he would have on a NL team. Examination also reveals that when substitute players are considered separately, the positive association between the number of blacks employed and winning is much weaker, whereas the negative association between black ballplayers and attendance is the same. Not coincidentally, black players made up a significantly smaller proportion of substitutes than they did of starters throughout the entire period.