This paper examines data on the racial composition and financial and sporting performance of professional English soccer teams between 1974 and 1993. In an earlier paper, Szymanski showed that teams with an above average proportion of black players would tend to perform better on average that would have been expected given the aggregate wage bills of these clubs. Since players are more or less freely traded in soccer this presents strong market-based evidence of discrimination. In the present paper we explore the source of such discrimination. In particular we are concerned to test the hypothesis that discrimination is attributable to the fans rather than the owners. If fans were racially prejudiced then the owners of a team might expect to generate a smaller marginal revenue product from a black player compared to an equally skilled white player. We assess the presence of fan discrimination by examining relationships between attendance, revenues, performance and the proportion of black players in the team. We also incorporate evidence regarding statements of racial prejudice (from the British Social Attitudes Survey) in particular regions. We find little evidence that the discrimination against black players has its source in fan discrimination.