BACKGROUND: The accuracy of restaging in patients with stage IIIa non-small cell lung cancer after neoadjuvant chemoradiotherapy is unknown. METHODS: A prospective trial of patients with biopsy-proven N2 disease who underwent initial clinical staging with mediastinoscopy, integrated positron emission tomography/computed tomography (PET/CT), and CT. Patients then were clinically restaged by the same imaging techniques 4 to 12 weeks after their induction chemoradiation therapy and then underwent definitive pathologic staging. RESULTS: Ninety-three patients had their lymph nodes pathologically restaged. Repeat PET/CT after neoadjuvant therapy missed residual N2 disease in 13/65 (20%) patients and falsely suggested it in 7 of 28 (25%). It was more accurate than repeat CT for restaging at all pathologic stages (stage 0, 92% vs 39%, P = .03; and stage I 89% vs 36%, P = .04). When the maximum standardized uptake value of the primary tumor is decreased by 75% or more, it is highly likely (likelihood ratio, +LR, 6.1) the patient is a complete responder; when it decreased by 55% or more, it is highly likely (+LR, 9.1) the patient is a partial responder. When the maximum standardized uptake value of the N2 node initially involved with metastatic cancer is decreased by more than 50%, it is highly likely (+LR, 7.9) the node is now benign. CONCLUSION: Repeat integrated PET/CT is superior to repeat CT for the restaging of patients with stage IIIa non-small cell lung cancer. The percent decrease in the maximum standardized uptake value of the primary and of the involved lymph node is predictive of pathology; however, nodal biopsies are required since a persistently high maximum standardized uptake value does not equate to residual cancer.