The Trier Social Stress Test (TSST) induces stress in the lab by having participants complete challenging tasks in front of an audience. The TSST has been adapted to virtual reality (VR), eliminating in-person audience variability that could explain sex differences in stress responses. Furthermore, the VR-TSST facilitates the examination of factors difficult to investigate in person, such as the effect of a large, 200-person audience. This study compared male and female physiological and psychological responses to an in-person 2-person TSST and a prerecorded VRTSST with audience sizes of 2 and 200 persons. Results indicated that only males had statistically significant cortisol reactivity to the TSST and responded with more positive affect, arousal, and lower ratings of stress than females. In the VR conditions these differences were less apparent, suggesting that the sex differences may be a result, at least in part, of in-person audience variability.