Objectives: This study examines the perceived effectiveness of bystander intervention in sexual harassment using virtual reality. Methods: A 2×3 factorial design compared two vignette presentation methods (VR vs. written scenarios) across three intervention conditions (direct, indirect, no intervention). The sample comprised 160 participants from a UK university. Outcome measures included self-report questionnaires capturing perception and emotion, alongside physiological sensing of electrodermal activity and heart rate. Results: Bystander action was noticed by victims, though type of action (direct or indirect) did not significantly influence immediate perceptions or emotions; both were reported as helpful. Direct intervention was more visible, but perceived as intrusive by some, while others found indirect intervention sufficient. Delivery method affected presence and therefore affective responses. Conclusions: Direct or indirect bystander intervention can be perceived as helpful. VR is a useful tool for assessing bystander intervention, but its novelty may amplify emotional responses, complicating the teasing apart of effects across different experiment conditions.