Background: Virtual reality (VR) is increasingly used to study and treat psychiatric disorders. Its fidelity depends in part on the extent to which the VR environment provides a convincing simulation, for example whether a putatively stressful VR situation actually produces a stress response. Methods: We studied the stress response in 28 healthy men exposed either to a stressor VR elevator (which simulated travelling up the outside of a tall building and culminated in the participant being asked to step off the elevator platform), or to a control elevator. We measured psychological and physiological (salivary cortisol and alpha-amylase, blood pressure, pulse, skin conductance) stress indices. We also measured subsequent performance on the N-back task because acute stress has been reported to impact on working memory. Results: Compared to participants in the control elevator, those in the external elevator had increases in skin conductance, pulse and subjective stress and anxiety ratings, altered heart rate variability, and a delayed rise in cortisol. N-back performance was unaffected. Conclusions: A putatively stressful VR elevator produces a physiological as well as a psychological stress response, supporting its use in the investigation and treatment of stress-related disorders, and its potential value as an experimental laboratory stressor.