Research on stereotype threat has repeatedly demonstrated that the intellectual performance of social groups is particularly sensitive to the situational context in which tests are usually administered. In the present experiment, an adaptation of the Raven Advanced Progressive Matrices Test was introduced as a measure of cognitive ability. Results showed that individuals targeted by a reputation of intellectual inferiority scored lower on the test than did other people. However, when the identical test was not presented as a measure of cognitive ability, the achievement gap between the target and the control group disappeared. Using heart rate variability indices to assess mental workload, our findings showed that the situational salience of a reputation of lower ability undermined intellectual performance by triggering a disruptive mental load. Our results indicate that group differences in cognitive ability scores can reflect different situational burdens and not necessarily actual differences in cognitive ability.