38 students enrolled in an introductory special education class in education of the emotionally disturbed were randomly assigned to two treatment conditions, a normal (control) condition and low expectancy condition. These teacher trainees participated in a two phase study. During phase 1 the teachers were asked to rate a hypothetical normal child (control group) or a hypothetical emotionally disturbed child (low expectancy group) on two dependent measures developed for this research. During phase 2, both groups independently viewed a videotape of the same normal child. The control group was told the child was normal; the low expectancy group was told the child was emotionally disturbed. Both groups completed the same dependent measures following observation of the child. Differences between the groups in both phases indicate that teacher trainees hold negative stereotypical expectations of children labeled emotionally disturbed. Observations of normal behavior alter these expectations to some extent, but the negative halo of the label still results in more negative perceptions of behavior than when the child is labeled normal.