This experiment used a physiological measure reflective of cognitive demand, heart rate variability (HRV), to examine performance disruption under stereotype threat. Both Science and Psychology majors in France (for which there are stereotypes suggesting higher intelligence among the former) were asked to complete a test that was described either as a valid measure of general intellectual ability (stereotype threat for Psychology majors) or as nondiagnostic of ability (control). An electrocardiogram measured HRV while students took the test. Performance on the test was lower for the Psychology compared with the Science majors, but only when the test had been described as diagnostic of intelligence. HRV data showed that both sets of students showed lower HRV in the diagnostic test condition, indicating increased mental load under evaluative scrutiny. However, decreased HRV was associated with poorer performance in the stereotyped group, but marginally higher performance in the non-stereotyped group. Stereotype threat appears to increase mental workload, but that increased workload appears to undermine performance only for those belonging to groups who are expected to underperform in a domain.