Clarke and colleagues1 are to be congratulated for their recently reported retrospective analysis of whether surgery is a prognostic factor for progression-free survival (PFS) based upon 2 datasets of patients (1 each in the pre- and post-temozolomide eras) treated according to the North American Brain Tumor Consortium phase II trials in recurrent glioblastoma (GB). We would like to make several comments regarding the report and the role of resective (nonbiopsy) surgery in the management of patients with recurrent GB. Counterintuitive results were reported, as surgery preceding enrollment in a clinical trial for recurrent GB did not affect the primary outcome of the analysis, which was 6-month PFS or overall survival. It is generally believed as a tenet of neuro-oncology treatment that reoperation for patients with recurrent GB is of value and positively influences outcome both by decreasing mass and attendant side effects and by improving results of subsequent therapy (either radiotherapy or chemotherapy) by diminishing tumor burden. The analysis of Clarke et al, however, suggests no difference in outcome in patients undergoing surgery, and consequently patients with smaller tumor masses at time of enrollment (the postsurgery cohort) in a clinical trial fared no better than patients not otherwise undergoing reoperation and with presumably larger tumors at the time of salvage treatment. Not clear from the analysis is whether patients undergoing surgery did indeed have smaller tumors than patients not undergoing reoperation. There was no re-analysis of tumor volume as defined by magnetic resonance imaging prior to study entry, so the statement that patients undergoing re-resection had smaller tumors is conjectural. Also not clear from the analysis was how a decision to undertake repeat surgery was determined. At present there are no clear guidelines for reoperation in patients with GB aside from the belief that reoperation is of value (diminishing tumor burden as well as decreasing side effects related to intracranial mass) and favorably influences outcome as defined by PFS and overall survival, a view that appears to be at odds with the Clarke analysis. Park and colleagues, in a recent provocative study,2 suggested 3 criteria for determining best candidates for reoperation in patients with recurrent GB: location in the brain (eloquent or non-eloquent), performance status (independent or otherwise), and expected survival ( 3 months). In most neuro-oncology centers, reoperation is driven primarily by tumor location, resectability, and performance and secondarily by symptoms related to tumor mass and the age of the patient. These less codified criteria are similar to those reported by Park et al but lack outcome data such as those provided by the Park researchers suggesting that there are clear candidates for which reoperation may have limited value. Importantly, it is unclear what criteria were used to suggest re-resection in the study by Clarke et al. How then to interpret this retrospective analysis? And does it affect present management of recurrent GB? Clarke et al1 conclude that surgery continues to offer benefit by reducing tumor mass–related symptoms, although no data were provided suggesting how many of the repeat surgery cohort in fact underwent surgery due to symptomatic tumor. In addition, repeat surgery may pathologically distinguish recurrent GB from treatment-related side effects such as pseudoprogression and radiation necrosis. Again, there are no data provided in the analysis by Clarke et al to indicate how often repeat surgery was used to make this clinically relevant distinction. Currently, there are no formal neuropathological criteria for grading previously treated gliomas. Because glioma grading was based on the most recent surgery, tumors that were lower grade at the initial surgery may have been upgraded and included in the GB group for further analysis. Tumors in the group that did not undergo reoperation were GB at the initial surgery. This may be an additional source of bias. Lastly, Clarke et al1 opine that repeat surgery allowed patients to enter a clinical trial that may have not been possible due to size of tumor and performance status. However, there are no data to support that the surgical cohort underwent surgery and converted from ineligible to eligible for a clinical trial. Furthermore, there are no data about preoperative and postoperative tumor volumes, or any comparison of volumes in the surgical and nonsurgical cohorts. A more provocative, and perhaps counterintuitive, conclusion based on the available data would be that repeat surgery does not impact survival, and consequently the value of repeat surgery in patients with recurrent GB otherwise eligible for a salvage clinical trial is poorly defined and perhaps minimal. As this conclusion is supported by the data presented, perhaps a re-analysis of the database, with inclusion of omitted information, such as surgical criteria, tumor volumes, alterations of performance status by surgery, extent of resection, and change in initial tumor grade, would help to determine whether or not this conclusion is valid.