Sixty-six agoraphobic patients were followed up five to nine years after their treatment in three clinical trials of behaviour therapy. The main outcome measures used in the original trials were repeated by an assessor who interviewed the patients. Ninety-five per cent of patients were interviewed and partial information was obtained on a further two patients. The measures taken at follow-up were compared with those obtained prior to treatment and six months after treatment ended. On most measures of agoraphobia the patients were much better at follow-up than they had been before treatment. The assessor's ratings suggested that there had been little change in the patients' agoraphobia since six months after treatment. Some of the patients' self-ratings showed evidence of a slight improvement over this period. No evidence of symptom substitution was found.