Purpose: Previous studies suggested an increased diagnostic yield for bronchoscopic (FB) transbronchial needle aspirate (TBNA) specimens from mediastinal lymph nodes when using rapid on-site evaluation by cytopathology (ROSE) but were limited by lack of randomization. The investigators performed the first randomized-prospective trial comparing ROSE(R) to no on-site cytopathology assessment (NR). Methods: All patients referred were eligible. 78 patients were randomized to R or NR groups. For R procedures, further specimens were deferred until results were available from the on-site cytopathologist. Diagnoses made by all procedures were recorded. The primary end-points were specimen adequacy and diagnostic yield.