Background. Interpersonal violence has increased as a health concern, especially in psychiatry practice, over the last decades. Nevertheless, most patients with stable mental disorders do not present an increased risk of violence, and mental disorder is not a necessary or sufficient cause of violent behaviours. People with mental disorders endorse more often a number of risk factors for violence that could confound this association, such as young age and male gender. The aim of this study was to investigate the effect of age, gender, and diagnosis on reported levels of interpersonal violence in a sample of people with severe mental illness. Methods. The sample was composed of 160 inpatients: 73 with a psychosis within the schizophrenia spectrum, 53 with a mood disorder and 34 with a personality disorder. All patients enrolled in the study were assessed for experiences of victimization and perpetration of interpersonal violence using the Karolinska Interpersonal Violence Scale interview. Demographic variables were also collected. Results. Both violence perpetration and victimization negatively correlated with age. Compared to males, females were exposed to higher degree of victimization in childhood and adulthood, whereas males were more involved in the perpetration of violence in childhood. Personality disorders were associated with higher levels of interpersonal violence, both perpetration and victimization; an interaction effect of gender and diagnosis was also observed for violence perpetration in adulthood. Distinct patterns of interpersonal violence did emerge for the diagnostic groups with mood disorder showing a victimization pattern, personality disorders a perpetration pattern and psychoses less defined patterns. Conclusions. The main finding is that psychotic disorders, mood disorders and personality disorders have different patterns of violent experiences interacting with age and gender. This study offers a better understanding of how gender and age could affect violent behaviours. Moreover, study findings may increase the comprehension of the reason why some mental disorders, compared to others, are more associated with the risk of victimization or perpetration of violence. These patterns could have pathophysiological or pathoplastic meaning addressing clinical and diagnostic trajectories and they could interact with other intervening risk factors.