In light of the poor reliability and discriminant validity of the DSM-III-R criteria for generalized anxiety disorder (GAD), extensive modifications were implemented in the development of the DSM-IV criteria. This study compares the discriminant validity of the DSM-III-R and DSM-IV criteria for GAD using data from a study of the familial aggregation of anxiety disorders and alcoholism. Based on information from a semi-structured diagnostic interview, both the DSM-III-R and DSM-IV associated symptom criteria were applied to characterize directly interviewed spouses and relatives of probands. The criterion-related and discriminant validity of the DSM-IV revised criteria were assessed according to the following clinical criteria: lifetime and concurrent comorbidity with depressive disorder and lifetime comorbidity with panic disorder. Subjects who met the DSM-IV associated symptom with and without comorbid depression were then compared on rates of treatment, psychotropic medication use, impairment, age at onset and presence of a psychosocial stressor at onset. No differences were found in the discriminant validity of the DSM-III-R versus DSM-IV definitions of GAD. It is postulated that the lack of difference in validity between the two diagnostic systems is due to the low prevalence of autonomic hyperactivity symptoms associated with GAD. Implications for the nosology of GAD and its relationship with other anxiety disorders are discussed. Copyright © 1999 Whurr Publishers Ltd.