OBJECTIVE: Psychosis is considered an important risk factor for violence, but studies show inconsistent results. The mechanism through which psychotic disorders influence violence also remains uncertain. The authors investigated whether psychosis increased the risk of violent behavior among released prisoners and whether treatment reduced this risk. They also explored whether active symptoms of psychosis at the time of violent behavior explained associations between untreated psychosis and violence. METHOD: The U.K. Prisoner Cohort Study is a prospective longitudinal study of prisoners followed up in the community after release. Adult male and female offenders serving sentences of 2 or more years for a sexual or violent offense were classified into four groups: no psychosis (N=742), schizophrenia (N=94), delusional disorder (N=29), and drug-induced psychosis (N=102). Symptoms of psychosis, including hallucinations, thought insertion, strange experiences, and delusions of persecution, were measured before and after release. Information on violence between release and follow-up was collected through self-report and police records. RESULTS: Schizophrenia was associated with violence but only in the absence of treatment (odds ratio=3.76, 95% CI=1.39-10.19). Untreated schizophrenia was associated with the emergence of persecutory delusions at follow-up (odds ratio=3.52, 95% CI=1.18-10.52), which were associated with violence (odds ratio=3.68, 95% CI=2.44-5.55). The mediating effects of persecutory delusions were confirmed in mediation analyses (β=0.02, 95% CI=0.01-0.04). CONCLUSIONS: The results indicate that the emergence of persecutory delusions in untreated schizophrenia explains violent behavior. Maintaining psychiatric treatment after release can substantially reduce violent recidivism among prisoners with schizophrenia. Better screening and treatment of prisoners is therefore essential to prevent violence.