Objective: We examined the possibility that maintenance cognitive behavior therapy (M-CBT) may improve the likelihood of sustained improvement and reduced relapse in a multi-site randomized controlled clinical trial of patients who met criteria for panic disorder with or without agoraphobia. Method: Participants were all patients (N = 379) who first began an open trial of acute-phase CBT. Patients completing and responding to acute-phase treatment were randomized to receive either 9 monthly sessions of M-CBT (n = 79) or assessment only (n = 78) and were then followed for an additional 12 months without treatment. Results: M-CBT produced significantly lower relapse rates (5.2%) and reduced work and social impairment compared to the assessment only condition (18.4%) at a 21-month follow-up. Multivariate Cox proportional hazards models showed that residual symptoms of agoraphobia at the end of acute-phase treatment were independently predictive of time to relapse during 21-month follow-up (hazards ratio = 1.15, p less than 0.01). Conclusions: M-CBT aimed at reinforcing acute treatment gains to prevent relapse and offset disorder recurrence may improve long-term outcome for panic disorder with and without agoraphobia. (Contains 1 table and 2 figures.)