In this study, we wished to compare the long‐term outcome of (medication‐free) panic disorder with agoraphobia patients randomized to cognitive or guided mastery therapy. Thirty‐one (67.4%) of 46 patients who had completed treatment were followed up about 18 years after end of treatment. In the combined sample and using intent‐to‐follow‐up analyses, there were large within‐group effect sizes of −1.79 and −1.63 on the primary interview‐based and self‐report outcome measures of avoidance of situations when alone, and 56.5% no longer had a panic disorder and/or agoraphobia diagnosis. No outcome differences between the two treatments emerged. Guided mastery was associated with greater beneficial changes in catastrophic beliefs and self‐efficacy. For two of five outcome measures, more reduction in panic‐related beliefs about physical and mental catastrophes from pre‐ to post‐treatment predicted lower level of anxiety from post‐treatment to 18‐year follow‐up when the effect of treatment changes in (a) self‐efficacy and (b) anxiety was controlled. However, for one of the outcome measures, this effect attenuated with time. Copyright © 2014 John Wiley & Sons, Ltd. Key Practitioner Message The results suggest that the very‐long‐term outcome of both cognitive therapy and guided mastery therapy for agoraphobia is positive.The results support the role of catastrophic beliefs as mediator of change.The pattern of results suggests that learning processes other than catastrophic beliefs may be important for long‐term outcome as well.