The diagnosis of generalized anxiety disorder (GAD) has been controversial since its inception. It remains unclear whether more stringent diagnostic crite- ria, such as in DSM-III-R, have improved the validity of GAD. Family studies suggest that GAD aggregates at least weakly in families of probands with GAD, and support the separation of panic disorder (PO) and GAD. Therefore, we can use a family study design to examine the validity of GAD. Independent familial transmission of GAD supports the validity of GAD. We report here the risk of GAD according to RDC, DSM-III, and DSM-III-R meteria in the first-degree relatives of probands @om four diagnostic groups: panic disorder, panic disorder with major depression, early-onset major depression (MDD), and normal controls. We did not find an elevated risk of DSM-111 or DSM-111-R GAD in the relatives of any of the ill proband groups compared to the relatives of the never mentally ill when controlling for proband comorbidity for GAD. I n contrast, RDC GAD aggregates in the first-degree relatives of probandsporn both of the PD proband groups (with and without MDD) compared to relatives of the normal control group. The inclusion of cases of subsyndromal panic attacks that did not meet the strict RDC for panic disorder as meeting the less restrictive RDC for GAD may partially account for the familial aggregation of RDC panic disorder and RDC GAD. RDC GAD seems to identijii) one or more syndrome(s) that may be on the familial spectrum of panic disorder. This syndrome may represent a mild or early variant of panic disorder. We also found a trend for RDC and DSM-111 GAD to aggregate in the $rst-degree relatives of the MDD proband group compared to the relatives of the never mentally ill controls. These data suggest that GAD demonstrates more independent familial transmission porn PD and MDD when defined by DSM-III-R criteria than when defined by RDC or and thus support the validity of DSM-111-R GAD. Anxiety 2:71-79 (1996). o 1996 ~~J I -L ~w , I ~C .