Objectives: Bystander studies have rarely considered the victimization risk associated with intervention into violent, dangerous emergencies. To address this gap, we aim to identify factors that influence bystanders’ risk of being physically victimized. Methods: We observed bystander behavior from video surveillance footage of naturally occurring violence in night-time economy settings, and data was analyzed with a logistic regression model. Results: Data shows that approximately one out of six interventions results in some type of victimization, typically with a relatively low degree of severity. The bystander’s social group membership, the setting of the emergency, and the bystander’s intervention type are estimated as risk factors for victimization. Conclusions: Previous research suggests that a bystander’s social group membership with victims promotes intervention behavior. Our results expand the role of social group membership as being a factor that also influences whether the intervening bystander is victimized.