Many violence prevention programs include a focus on the role of bystanders and third parties in violence prevention training. Central to this work has been the classic social psychological research on the “bystander effect”. However, recent research on bystander behavior shows that the bystander effect does not hold in violent or dangerous emergencies. Meta‐analyses of the literature show that the presence of others can facilitate as well as inhibit intervention in emergencies. Studies of real‐life bystander behavior captured on CCTV cameras shows that some bystander intervention is the norm and that the likelihood of bystanders being victimized is low. One reason for the limited effectiveness of violence reduction programs may be their approach to bystanders. We argue that violence reduction programs should: recognize that some intervention is likely (although it may not always be successful); see the group as a route to successful intervention rather than a threat to the likelihood of any single individual becoming an intervener; inform bystanders of the real risks of victimization; utilize the power of social relations between bystanders, victims, and perpetrators to enhance successful intervention; seek to deliver bystander intervention training in situ, rather than away from the context of the aggression or violence.