Objective: Evidence has emerged of a higher risk of serious violence in first-episode psychosis. This study compared the characteristics of two groups of persons with psychosis found not guilty of violent offenses on the grounds of mental illness (NGMI): those who had not yet been treated and those who had previously received treatment. Methods: Demographic, clinical, and offense data were taken from psychiatric reports and court documents for persons found NGMI of homicide, attempted homicide, and assaults causing serious injury in the state of New South Wales, Australia. Results: Between 1992 and 2008, 272 persons were found NGMI of homicide (N=138) or attempted homicide and assault resulting in serious injury (N=134). Schizophrenia-related psychosis was the most common diagnosis (N=234, 86%); 93 (34%) had a co-occurring substance use disorder. Of the 272 offenders, 124 (46%) were classified as being in their first episode of psychosis, including 72 (52%) of the homicide offenders and 52 (39%) of those who committed a nonlethal violent offense. Multiple logistic regression analysis found that compared with those who had received previous treatment for psychosis, the first-episode patients were younger, more likely to be from a non-English-speaking country, and more likely to have used a firearm in the offense. Conclusions: The first episode of psychosis is a period of particular risk for homicide and serious violence. Younger age, language and cultural barriers to care, and greater availability of lethal means appear to contribute to an increased risk of homicide and of other serious violent offenses in first-episode psychosis.