Data were obtained in two areas dealing with expectancy effects: When are tutors susceptible to expectancy effects and do expectancies inevitably lead to discrimination against low ability students? We arranged for 76 black undergraduates to teach a lesson dealing with Eire safety in the home to a 10‐year‐old boy. Two black males and two white males served as confederates. Subjects were told that a child had either high I.Q., low I.Q., or were given no information (the control group). Subjects' knowledge about the black confederates' alleged I.Q. did not influence expectancies or subsequent teaching behavior toward low I.Q. children. On the other hand, subjects rated white confederates in the high I.Q. condition as more intelligent than children in the low I.Q. condition. White confederates in the low I.Q. (or low expectancy) condition were asked more questions and given fewer expressive gestures than white confederates in the control condition. In addition, subjects maintained more physical distance with white than black confederates. The present results were compared to earlier research on white subjects dealing with nonverbal mediators of expectancy effects. It was suggested that an adequde model of expectancy effects must consider the contribution of cognitive mediators to understand when and how the phenomenon occurs.