INTRODUCTION: The use of psychoactive substances is associated with changes in emotions, memory and executive functions, while chronic use can lead to organic changes in various regions, including the prefrontal cortex. It is known that lesions in this area cause disinhibited and even aggressive behavior in humans. According to previous research, abuse of alcohol, cannabinoids and cocaine show the highest statistically significant correlation with violence. OBJECTIVES: To examine the connection between history of addiction or harmful use of psychoactive substances with psychological characteristics, features of offenses and criminal responsibility of offenders examined at the Institute for Forensic Psychiatry "Dr. Vlado Jukić". PARTICIPANTS AND METHODS: The research is based on retrospectively collected data. The participants are criminal offenders who underwent a psychiatric examination during the court process in the period from January 1, 2012 to December 31, 2021. The data includes histories of psychiatric illnesses, results of psychological testing of impulsivity, aggressiveness and dissocial/antisocial characteristics, type of criminal offense and types of psychoactive substances from the subject's history. RESULTS: In a sample of 192 participants, it was found that those who have a diagnosis of addiction or harmful use of psychoactive substances more often showed dissocial/antisocial features on tests and were most often of diminished criminal responsibility. A significant correlation was also found between the types of psychoactive substances used in persons with this diagnosis and the use of stimulants, as well as the presence of impulsivity. CONCLUSION: The research found significant differences in the use of stimulants, criminal responsibility and dissocial/antisocial characteristics between participants with a diagnosis of addiction or harmful use and those without it. The population with impulsivity shows a different pattern of psychoactive substances abuse than those without it. However, future research should include larger samples and perform more complex statistical analyses.