Objective: This study aims to determine whether it is possible to identify clinical profiles at risk of violent behaviors (VB) in the early phase of psychotic disorders, on the basis of the main dynamic psychopathological risk factors and describe characteristics of the groups with highest levels of violent behaviors. Method: A total of 265 patients, aged 18 to 35, treated at the Treatment and early Intervention in Psychosis Program (TIPP - Lausanne), a specialized early psychosis program, were included in this study. We conducted a latent-class analysis and a discriminative analysis on the basis of the main dynamic VB risk factors: substance use disorder, impulsivity, positive symptoms, insight, aggression, hostility, anger, emotional instability and adherence to treatment. These factors were evaluated by specialized scales and on the basis of the Positive and Negative Syndrome Scale (PANSS). VB were restricted to physical aggression against people, defined as “serious violence”. They were assessed on the basis of a questionnaire listing violent offenses (Swiss Criminal Code) and VB such as assault and battery; information through the forensic psychiatric services and on the basis of the Staff Observation Aggression Scale (SOAS-R scale) during inpatient treatment phase. Results: Four heterogeneous subgroups were identified with respect to the studied clinical characteristics, including two groups with high rates of VB. The first group, comprising 46% of patients with VB, is distinguished by the prevalence of a range of dimensions related to hostility, impulsivity and emotional instability, associated with high levels of substance abuse and positive symptoms. These clinical dimensions are very significant at the statistical level, since they explain 70% of the construction of subgroups (discriminant analysis). The second group with 37% of patients with VB, is characterized by a lack of insight, lack of adherence to treatment and substance use. These two clinical profiles could increase the impairment of cognitive, functional and relational abilities and contribute to the development of VB in this early phase of psychosis. The third subgroup, with a violent behaviors rate of 28.6%, is distinguished by its high proportion of diagnoses of substance abuse (100%) and women (54%). A last subgroup of patients, the largest quantitatively, has a low proportion of VB (15%) and the lowest levels on the studied factors, suggesting that the majority of patients with this profile commit few VB. Conclusion: Our results show that it is possible to identify groups at risk of violent behaviors during the early phase of psychosis, on the basis of clinical characteristics that may evolve and therefore be the focus of preventive care. These results highlight the need to target substance use, impulsivity and lack of insight at follow-up in order to prevent VB.