Studies have provided empirical evidence, that teachers' judgments of students' academic performance are biased by social stereotypes. Thus, in the present study it was tested whether the impact of stereotypes about immigrant students, students with a low socioeconomic status and male students on preservice teachers' evaluations can be reduced by a counterstereotypic training. For this purpose, one hundred and forty-seven preservice teachers first read a short text in which a fifth grader male student was described who could respectively be allocated to one of these three stereotypical categories. Half of the participants then completed the training. After that, all subjects were asked to evaluate an essay of the fictive target and to give a prognosis about his level of graduation. The results showed that the training led to a reduction of the stereotype-based judgment bias only with regard to the gender stereotype. Accordingly, in the training condition the male target's essay was marked better and the prognosis of his level of graduation was higher than in the no-training condition. No training effects were found for the other two stereotypes. The implications of the study are discussed with regard to teacher education and vocational trainings.