Objective: Termination in psychodynamic therapy (PDT) is a potentially conflictual and turbulent phase for patients, with a risk for increases in symptoms. However, few studies of PDT have assessed symptoms frequently enough during treatment to determine whether such setbacks are in fact common in PDT. Methods: In a doubly randomized clinical preference trial, 217 adults (female = 163; M age = 34.8 (12.6)) with panic disorder with or without agoraphobia were treated with panic-focused psychodynamic psychotherapy (PFPP) or panic control treatment (PCT), a form of cognitive behavioral therapy. Participants completed the Panic Disorder Severity Scale-Self Report (PDSS-SR) weekly during treatment (weeks 1-12), and 6, 12 and 24 months after treatment. Using piecewise latent growth curve modeling we tested the trajectories of change focusing on the termination phase in PFPP. Results: Week-to-week improvement on the PDSS-SR stopped (a termination setback [TS]) in PFPP during weeks 10-12, whereas PCT participants continued to improve. Larger symptom reductions up to week 10 in PFPP predicted a more severe TS. Less avoidant attachment and less severe interpersonal problems also predicted more severe TS. The TSs tended to last, as evidenced by inferior outcomes, up to the 12-month follow-up. Conclusions: This study provides evidence of a TS in PDT. Resurgence of symptoms as termination approached was more common in PFPP than in PCT. Studies involving weekly assessment of primary and comorbid symptoms, as well as qualitative analyses of the patient experiences of the therapeutic process during termination, in different forms of PDT, are warranted.