Four studies were conducted to examine whether cognitive appraisals, manipulated through task instructions, would moderate social-facilitation effects. In Study 1, participants in the challenge condition performed better on a mental arithmetic task when the experimenter was present. Conversely, participants in the threat condition performed worse when the experimenter was present. Study 2 extended these findings across 2 math tasks that varied in degree of difficulty. The pattern of performance data failed to support prior drive theories and provided support for a unique contribution of cognitive appraisals in explaining social-facilitation effects. Study 3 validated the appraisal manipulations by using multiple measures of cognitive appraisals. Finally, Study 4 offered increased validity by replicating the performance data using an anagram task.