This paper examines three issues concerning the frequently documented negative correlation between formal education and ethnic prejudice, namely its reliability, its validity and the manner in which it is mediated. Reliability is demonstrated across three indices of ethnic attitudes in seven representative samples drawn from four European countries (West Germany, Netherlands, France, Great Britain; total N=3788). The hypothesis that this correlation reflects only the tendency of more highly educated respondents to give more socially desirable answers and not true attitude differences was inconsistent with the finding from the survey data that educational level also correlated negatively with responses to an index of subtle prejudice. Results from an experiment employing the bogus pipeline procedure similarly refute this hypothesis, indicating that significant education-related differences in expressed prejudice remain under conditions in which the tendency to give socially desirable responses is reduced. Finally, path analysis based on the survey data show that part but not all of the association between low education and ethnic prejudice is mediated by social psychological variables, particularly group relative deprivation, perceived belief incongruency, political conservatism, and acceptance of inter-ethnic contact.