Background: Few clinical trials of cognitive behaviour therapy (CBT) for anxiety disorders have conducted follow-up beyond one year post-treatment. This paper summarises the long-term outcome of eight clinical trials of CBT for anxiety disorders in terms of diagnostic status, healthcare usage and symptom severity and compares the symptom profile of participants with the best and worst outcomes relative to chronic depression and the normal population. Methods: Follow-up at 2-14 years with 396 patients (51% of those available to contact) employed structured diagnostic interview, assessment of healthcare usage and self-report measures of symptom severity. This paper concerns 336 participants who had either no disorder or at least one anxiety disorder and information on healthcare usage over the follow-up period. Results: Only 38% recovered with little or no treatment over the follow-up period while 30% had a very poor outcome despite extensive treatment for anxiety over many years. The symptom profile of this 'treatment-resistant' group was comparable to 76 patients with chronic depression and significantly worse than normative data for psychiatric outpatients. Chronic anxiety disorder with co-morbid depression has a more severe symptom profile than chronic anxiety disorder alone. Limitations: The follow-up sample, although broadly representative, may have a bias towards a more favourable picture of overall outcome. Conclusions: The long-term outcome of anxiety disorders, irrespective of diagnosis or active treatment, is diverse but with a tendency towards chronicity. Distinctions between acute and chronic presentations of common mental disorders are more important than distinctions between chronic anxiety and chronic depression.