Background Bullying is a prevalent phenomenon that can have an array of negative impacts on both victims’ and perpetrators' long-term health and wellbeing. Despite the widespread assumption that empathy should be a key target for anti-bullying interventions, research examining this relationship is surprisingly scarce, and the evidence base informing the implementation of empathy-focused interventions remains in its infancy. Using data from the Stand Together trial, we examined how affective and cognitive empathy predict the status of victim, bully, and bully-victim. Methods We used a longitudinal trial with data collected in 2021 and 2022 from 4660 UK primary school children aged 6-11 years, including measures of empathy, victimisation, and involvement in bullying. We used propensity score matching and multinomial logistic regression to explore how children’s self-reported empathy towards victims of bullying at baseline predicted their role in bullying at one-year follow-up. Results Consistent with existing literature, we found that low affective empathy was a significant predictor of bullying perpetration at follow-up (OR = 0.60, 95% CI [0.45, 0.81], p < .001), but so was low cognitive empathy (OR = 0.73, 95% CI [0.56, 0.95], p < .001. We also found that both high affective (OR = 1.34, 95% CI [1.23, 1.47], p < .001) and cognitive (OR = 1.33, 95% CI [1.23, 1.44], p < 0.001) empathy predicted later victimisation. Conclusion The findings identify high empathy as a new risk factor for peer victimisation and confirm the role of low empathy as a predictor of later bullying perpetration. We discuss how these findings can inform the strategic integration of empathy training to enhance the effectiveness of bullying prevention efforts.