Objective: Accumulating evidence shows that bystanders witnessing public disputes frequently intervene to help. However, little is known regarding the risks entailed for those bystanders who enter the fray to stop conflicts. This study systematically examined the prevalence of bystander victimizations and the associated risk factors. Method: Data were a cross-national sample of 93 surveillance camera recordings of real-life public disputes, capturing the potential victimizations of 417 intervening and 636 nonintervening bystanders. Results: Data showed that interveners were rarely physically harmed-at a rate of 3.6%-and noninterveners were virtually never victimized. Confirmatory regression results showed that conflict party affiliation was a moderately robust predictor of bystander victimization. The gender of the intervener was a highly fragile risk factor. More severe conflicts were not associated with a higher victimization likelihood. Conclusions: Our findings highlight the value of naturalistic observation for bystander research and emphasize the need for evidence-based bystander intervention recommendations. Data, materials, and postprint are available at osf.io/vyutj.