Background Rapid on-site evaluation (ROSE) is a technique where transbronchial needle aspiration (TBNA) cytology samples are rapidly stained and screened for diagnostic material in the procedure room, during the procedure. We hypothesized the sensitivity of ROSE in patients with sarcoidosis is very low, leading to unjustified use of an expensive technique.Data and Methods This was a retrospective study at an inner-city hospital. Medical records of all patients who underwent EBUS-TBNA of mediastinal and hilar lymph nodes with ROSE over a 3-year period were evaluated. The sensitivity, specificity, and positive and negative predictive values of ROSE in patients with sarcoidosis were calculated, with pathologic diagnosis by cell block as the “gold standard.” Patients with malignancy were used as a comparison. Results One hundred eighty-four patients who had ROSE on EBUS-TBNA of mediastinal and hilar lymph nodes were included. Thirty were diagnosed with sarcoidosis, 95 with malignancy, and 59 with benign lymph nodes. The sensitivity of ROSE in patients with sarcoidosis was 44%, specificity and positive predictive value were 100%, and negative predictive value was only 17%. Conclusion Given low sensitivity and negative predictive value, ROSE may not be as useful in diagnosing sarcoidosis as it is in diagnosing malignancy.