Previous studies, using data from the 1980s, found that racial composition of NBA teams is positively correlated with racial composition of the metropolitan markets in which the teams are located. Researchers have interpreted this evidence as consistent with a "customer discrimination" hypothesis. We reconsider this hypothesis by examining evidence from the 1990s and generate three principal findings. First, based on player performance statistics, we find no evidence of discrimination at the league level--that is, the best players appear to be playing in the league regardless of race. Second, players, categorized by race, are not randomly distributed across teams. Instead, the relationship between team racial composition and metropolitan area racial composition, while weaker than in the 1980s, persists in the NBA in the 1990s. Hence, teams located in areas with greater concentration of white population may find it revenue enhancing to cater to customer demand for viewing teams that include white players. Our third finding, based on revenue from home game attendance, is that as the number of white players declined significantly over the decade, the revenue product of a white player increased on the margin. This effect appears to be more pronounced for teams located in cities with larger white populations. We also find evidence that, in recent years, the top-performing white players in the NBA tend to locate in cities with larger white populations, suggesting that teams in these cities place a higher marginal value on such players.