As noted in the commentary by Cesario et al. (1), an evidence-based consensus on the role of surgery in multimodality treatment for stage IIIB non-small cell lung cancer (NSCLC) (excluding patients with pleural effusion) is indeed needed. These authors point out that several studies that have examined outcomes of patients with IIIB disease treated with neoadjuvant therapy followed by surgery, including our Phase I study, typically considered surgery only in patients with clinical downstaging from neoadjuvant therapy (2, 3). The best evidence for this approach is provided by the Phase II Southwest Oncology Group (SWOG) 8805 study, in which 126 patients with stage IIIA or B NSCLC received induction cisplatin and etoposide plus concurrent thoracic radiotherapy (2). Resection was attempted in patients without progressive disease. Fifty-one patients (40%) had stage IIIB (T4 primary or N3 nodes) disease. In this subset, resectability was 80%, and 3-year survival was 24%, similar to results for the stage IIIA group. The strongest predictor of survival after surgery was nodal status at the time of resection. In patients with pN0 disease at resection, 3-year survival was 44%, versus 18% in patients with involved nodes (P = 0.0005). The fact that this survival difference was seen in patients with both IIIA and IIIB disease suggests that treatment decisions should not only take anatomic staging into account but should also consider tumor biology and response to therapy. Thus, some patients with stage IIIB disease, for whom guidelines typically recommend combined chemoradiotherapy without surgical resection (4), may, by virtue of an excellent response to neoadjuvant treatment, be able to enjoy the benefits of resection, as would a patient with similarly responsive stage IIIA disease. The final report of the randomized intergroup trial 0139 (RTOG 93-09), investigating the outcomes of patients with stage IIIA NSCLC from combined chemoradiotherapy alone versus chemoradiotherapy followed by surgery, is pending (5). Preliminary data suggest similar overall survival in both groups of patients but improved progression-free survival in the surgical arm. Further analysis may shed light on who will benefit from the full trimodality approach to therapy for locally advanced NSCLC.Based on our results, in which 3 of 12 IIIB patients were deemed resectable, and the SWOG 8805 data, we agree with Cesario et al. (1) that surgical resection after neoadjuvant therapy can be considered in selected patients with stage IIIB (T4 primary or N3 nodes) after neoadjuvant therapy. The patient in question with T4 disease had radiographic evidence of invasion into the mediastinum on presentation, not a satellite nodule or carinal involvement that might meet the more typical guidelines for resectability. The question remains how to best identify patients who may benefit from surgery. In our Phase I study, all patients were restaged by computed tomography (CT) alone after both induction chemotherapy and chemoradiotherapy. Of the 14 patients eligible for surgery based on restaging CTs showing the equivalence of stage IIIA disease or better, 7 of the preoperative CT scans did not accurately predict the postoperative pathological stage. Of these, six patients were downstaged to N0 status or had a pathological complete response to neoadjuvant therapy, indicating that restaging by CT alone is not fully reliable.Alternative approaches to restaging include positron emission tomography (PET) and repeat mediastinoscopy. PET restaging is attractive because it offers the possibility of both characterizing locoregional response to neoadjuvant therapy and identifying clinically unrecognized distant metastatic disease, thereby sparing those patients unnecessary treatment (6). Unfortunately, the data available to date include unacceptable numbers of false positives and false negatives for PET restaging alone to be considered a standard approach to determining resectability of stage III NSCLC after neoadjuvant therapy (7, 8). Repeat mediastinoscopy, as Cesario et al. (1) have observed, can be technically challenging in the postchemoradiotherapy setting and is therefore not routinely performed. As a result, it is difficult to recommend a standardized approach to restaging after neoadjuvant therapy in patients with stage IIIB NSCLC to determine resectability. Nevertheless, the favorable outcomes that can be seen in selected stage IIIB NSCLC patients in response to trimodality therapy do indicate that further study in identifying appropriate surgical candidates is clearly needed.