Previous research suggests that a supervisor's liking of a subordinate biases evaluations of that subordinate's performance. Other research suggests that supervisor liking influences the treatment of the subordinate, which may in turn influence the subordinate's performance. The present study investigated these influences in a laboratory simulation. Subjects (N=140) worked on their own tasks and supervised an alleged subordinate during a 30-min work period. Supervisor liking of a subordinate was manipulated by altering the personality characteristics and the attitude similarity of the subordinate. Reward context was manipulated by rewarding supervisors for either the total performance of both the supervisors and subordinates or for the individual performance of each supervisor only. Supervisor liking positively influenced the expected leader-member exchange, treatment of the subordinate, and evaluations of subordinate performance. The reward context influenced supervisor treatment of the subordinate and the number of problems the supervisor solved.