Background: The construct of aggression is central to work with violent offenders, but it is a broad construct that can be assessed by many different methods and instruments. Its measurement may, however, have profound implications for treatment planning. We need more knowledge about how different methods for assessing aggression relate to each other. Aims: Our aims were to investigate, first, the convergence and concordance of two methods of assessing aggression: self-report and clinical assessment and, second, to determine the degree to which aggression can be discriminated from neighbouring constructs, such as hostility, anger and criminal behaviour. Methods: A nationally representative Swedish cohort of 269 18–25-year-old incarcerated violent offenders was recruited. Data were collected through structured self-reports of aggression, anger and hostility traits (Aggression Questionnaire-Revised Swedish Version) and clinical assessments of lifetime prevalence of aggressive and antisocial behaviours (Life History of Aggression). Criminal records were retrieved from the Swedish National Crime Register. Results: Self-ratings and clinician-ratings of aggression were highly convergent and concordant, especially regarding physical aggression. Violent offence records were weakly, if at all, correlated, while self-reported hostility was weakly, or not at all, correlated with self-reported or with clinician-rated aggression. There was an inverse relationship between aggression and criminal records of sexual offences. Conclusions and Implications: Even though a combination of self-reports and clinician-ratings may provide a better overview of an individual's aggressive behaviours, our results indicate that they yield such similar information that either alone would be sensitive enough. Our results do not, however, support using one of these methods as a proxy for the other since choice of measure and accepted concordance between them depend on the context within which the assessment is conducted. We reconfirmed that official records of violent offending are unlikely to be adequate measures of outcome after interventions to reduce aggressive behaviours.