One hundred and twenty six undergraduates were cast into a supervisory role in which they worked with a subordinate (confederate) for two twenty minute work sessions. Subjects were placed into one of nine conditions. Subordinate reaction (positive, negative, or none) and subsequent performance (increased, decreased, or same) were manipulated by the confederate. Both reaction conditions were predicted to influence a leniency bias in the supervisor's rating behavior, and either increase (negative reaction), or decrease (positive reaction) amount of supervision. Also, supervisors exposed to the positive reaction were hypothesized to provide more self attributions, while supervisors in the negative reaction group should tend toward greater self-serving attributions. In addition, supervisor response to interpersonal attraction, conflict avoidance, and uncertainty scales were obtained. Both ratings and amount of supervision measures failed to yield significant results. Partial support was found for the attribution hypothesis, and differences on the interpersonal attraction scales were obtained for the experimental reaction conditions. Implications of the results and suggestions for possible research are discussed.