OBJECTIVE: The authors review the empirical data on generalized anxiety disorder, a diagnostic category that has been among the more conceptually challenging in psychiatric nosology. METHOD: Published studies and recent findings that were considered by the Generalized Anxiety Disorder Subcommittee of the DSM-IV Anxiety Disorders Work Group are reviewed. Among the issues examined are diagnostic reliability, comorbidity, boundaries with other disorders, and clinical features. RESULTS: A variety of data on the reliability and validity of generalized anxiety disorder have been produced. Some authors have suggested that generalized anxiety disorder is better conceptualized as a vulnerability that should be located on axis II, and others have recommended that the category be eliminated. Although the diagnostic reliability of generalized anxiety disorder is lower than that of other anxiety disorders, the features constituting the diagnostic criteria for generalized anxiety disorder have been found to be reliable. An important development has been the determination of a set of somatic symptoms associated with generalized anxiety disorder that differs substantially from those for other anxiety disorders. These findings led to reduction in the number of items in the symptom criterion, from 18 in DSM-III-R to six in DSM-IV. Another substantial revision is greater emphasis on the uncontrollability of worry. CONCLUSIONS: Whereas the data on construct and discriminant validity, age at onset, course, familial transmission, and response to treatment generally support the DSM-IV definition of generalized anxiety disorder, the construct continues to have weaknesses and further research is needed.