In this issue of CHEST (see page 491), Kates et al1 report their analysis of Surveillance, Epidemiology, and End Results (SEER) registry data in patients with stage IA non-small cell lung cancer (NSCLC) treated either with limited resection or lobectomy. Conclusions from this analysis indicated that there was no difference in outcomes between the two groups, which included 2,090 cases of stage I NSCLC ≤ 1 cm. By performing Cox proportional hazard analysis adjusted for treatment propensity, the investigators have attempted to provide some sense of balance between the limited resection and the lobectomy groups. The number of patients in this and other similar studies is large and cannot be dismissed. This well done study by Kates et al will add fuel to the lobectomy vs limited resection debate. However, in reality, the SEER database is simply a repository of voluntary data submissions, and propensity scoring does not address the major limitation of these types of data sets, namely selection bias by the surgeon. The SEER database offers no information regarding preoperative pulmonary function tests, blood gases values, or activity level that could be included in the analysis. Although some would argue that limited resection is performed on a sicker subset of patients who are unable to tolerate lobectomy, the SEER database does not have sufficient data to address this issue. We, as surgeons and pulmonologists, are left to determine how this study will affect the future treatment of patients with stage I lung cancer.