Cyberbullying behavior (CB) on social media is complex because its perpetrators exhibit varied demographic characteristics and personalities. Prior studies have applied Big Five (Big5) and Dark Tetrads (Dark4) personality traits (PTs) along with demographic attributes, using symmetrical modelling, but revealed mixed and inconsistent results. This paper applies an asymmetric modelling approach using complexity and configurational theories to develop configurations of PTs and demography to predict CB. The online survey data have been analyzed using fuzzy set qualitative comparative analysis (fsQCA) technique. Regarding Big5 PTs, our findings reveal that, for instance, people scoring high in conscientiousness, neuroticism, openness and low in agreeableness undertake cyberbullying. For Dark4 PTs, the combination of either psychopathy and sadism or Machiavellianism and psychopathy leads to cyberbullying. As for demographic attributers, educated young married people, irrespective of gender, are likely to commit cyberbullying. Our all-inclusive model reveals that social media bullies, regardless of their gender, marital status, and social media experience, are young, educated, neurotic, conscientious, psychopathic, and sadistic with high Machiavellianism and low agreeableness. Further, we suggest configurations to reduce cyberbullying. The findings are discussed with implications for theory and practice.