The psychometric tools used for the assessment of generalized anxiety disorder (GAD) either do not conform to the current concept of the condition or have important limitations. We aimed to develop and validate a new questionnaire for the assessment of symptom profile and severity of GAD. An original pool of potential scale items was subjected to a series of studies in non-clinical and clinical populations, in order to determine the final composition of the scale. The psychometric properties of the new scale, the Generalized Anxiety Disorder Inventory (GADI), were evaluated using a factor analytic model suitable for ordinal data and the Graded Response Model. The precision of measurement of the GADI was quantified through the item information functions. A total of 197 outpatients and 522 non-clinical subjects participated in four studies and completed the GADI. The final 18-item scale was derived from an original pool of 30 potential items. The GADI showed good reliability, convergent and divergent validity. The scale comprises three factors, relating to cognitive, somatic and sleep symptoms. It accurately distinguished GAD patients from non-patient controls. The cognitive factor also distinguished GAD from other anxiety disorders and depression. The GADI is a useful tool in the assessment of the breadth of symptoms and the severity of generalized anxiety disorder in clinical settings.