Currently available symptom severity measures for Generalized Anxiety Disorder (GAD) are not optimal. This study investigates the reliability and validity of a new measure for GAD. The Generalized Anxiety Disorder Severity Scale (DGSS), comprising 8 DSM-IV GAD symptoms assessed in terms of frequency and intensity, was used in a trial of agomelatine versus placebo for the treatment of GAD. Internal reliability, concurrent validity, responsiveness to change, most robust items, and factor structure were computed. The DGSS demonstrated good internal reliability, correlated significantly with the Hamilton Anxiety Scale and Clinical Global Impression severity scale, and demonstrated a clear change in response to agomelatine. The most robust DGSS items were derived, and an exploratory factor analysis yielded a 2-factor structure of the DGSS. The DGSS is potentially a useful scale for the assessment of GAD in clinical trials of this disorder.