The diagnostic stability and long-term prognosis of generalized anxiety disorder (GAD) remain the subjects of considerable controversy. We report the results of an investigation of the long-term outcome of an original sample of 44 subjects who participated in a medication trial. Subjects were reinterviewed approximately 16 months after completion of the study, using structured interviews. Fifty percent continued to fulfill criteria for GAD. Other concurrent axis I diagnoses were as follows: dysthymia, 11%; major depression, 7%; and social phobia, 7%. Regarding axis II comorbidity, subjects with chronic GAD were more likely to fulfill criteria for one or more personality disorders, especially in clusters B and C. In addition, follow-up subjects with GAD and with remitted GAD reported a statistically equivalent number of recent life events, although subjects with chronic GAD were more likely to report significant dissatisfaction with life. The findings indicate that although many subjects with GAD do not follow a chronic course, many others remain symptomatic. The results also suggest that GAD symptoms are not simply the result of a subject's recent negative experiences, and that life satisfaction measures are an accurate reflection of GAD outcome.