ABSTRACT Background Paranoia is a common experience prevalent in the general population. Social identity refers to our sense of belonging to a social group and has been implicated in the formation and maintenance of paranoia. Research into these mechanisms is still emerging. It was hypothesised that trust and hostile attribution bias would mediate this relationship in a UK general population sample. Methods A cross-sectional online survey was conducted with 355 UK residents. Measures of family and friendship social identity, trust, hostile attribution bias, paranoia, and psychosis proneness were completed. Results A linear regression found that social identity significantly predicted paranoia, and unexpectedly, this was replicated for psychosis proneness. This was a negative association whereby high social identity scores predicted lower paranoia and psychosis proneness scores. A parallel mediation model indicated family and friendship group identity was associated with lower paranoia and lower psychosis proneness when participants reported higher levels of trust and lower levels of hostile attribution bias. Discussion Social identity is associated with paranoia and psychosis proneness, and these effects are mediated through trust and hostile attribution bias. The findings have implication for targeting research and interventions on social group membership.