The psychometric properties of a Dutch version of Buss and Perry's Aggression Questionnaire (AQ) were examined in a sample of violent forensic psychiatric inpatients and outpatients and a sample of secondary vocational students. The internal consistency, interitem correlations, and item-scale correlations of the subscales Physical Aggression, Anger, and Hostility were good but turned out to be unsatisfactory for the subscale Verbal Aggression. The four-factor structure of the AQ could not be confirmed, but the four-factor structure of a 12-item version (short form) of the AQ, the AQ-SF, produced an acceptable fit. The test-retest reliability of the AQ and AQ-SF total scores was good, although the test-retest reliability of the AQ-SF subscale Physical Aggression was not satisfactory. The validity of both the AQ and AQ-SF could be demonstrated by meaningful correlations with alternative measures of aggression and personality, but inpatients were not found to display higher scores on the AQ or AQ-SF than the students.