We performed a retrospective review of all mediastinal and hilar tissue samples submitted for pathologic analysis at our institution over an 8.4-year period (January 1, 1997, to May 15, 2005). The study was approved by the institutional review board at the University of Chicago with a waiver of consent since the study involved an anonymous, retrospective evaluation of medical records only. One thousand six hundred six billing records were queried for all patients who underwent bronchoscopy with TBNA. In addition, 2,848 pathology records were queried for the following terms: “mediastinum”; “Wang”; “transbronchial needle aspiration”; “lymph node”; and “mediastinoscopy.” These two databases were merged to yield a total of 2,543 patients, which included all patients who had intrathoracic LNs sampled by bronchoscopic and surgical approaches. Because endoscopic ultrasound (EUS) and CT scan-guided biopsies accounted for a very small fraction of the intrathoracic LNs sampled (< 3%), they were not included in the analysis. The age, gender, procedure date, mediastinal/hilar sampling approach, LN stations biopsied, and biopsy results were recorded for all patients. Biopsy results were categorized as follows: (1) malignant; (2) benign pathology (eg, granuloma); (3) normal lymphoid tissue; and (4) nondiagnostic (eg, no tissue/cells, blood only, or bronchial mucosa cells only). Intrathoracic LN sampling success was defined a priori as successful sampling of any mediastinal or hilar tissue (malignant pathology, benign pathology [eg, granuloma], or normal lymphoid tissue).