Using a factorial survey experiment, we investigate whether teachers discriminate against lower class and minority children. The vignettes include information on the pupil’s gender, social and ethnic origin, academic potential, motivation, behaviour in class as well as parental educational aspirations. While the discrimination hypotheses are not supported, we find gender-specific effects of challenging behaviour, to the disadvantage of girls. Both the usefulness as well as potential pitfalls of the factorial survey approach are discussed.