BackgroundEndobronchial ultrasound‐guided transbronchial needle aspiration (EBUS‐TBNA) is routinely performed to confirm a lung cancer diagnosis and/or to clinically stage disease. EBUS‐TBNA findings may be used to determine whether patients can be offered potentially curative surgery. In this study, we evaluated the reporting in our service of EBUS‐TBNA cytology for early‐stage (operable) non‐small cell lung cancer (NSCLC), focusing on diagnostic accuracy and analyzing cases with discordant cytologic and post‐surgical histopathologic conclusions.MethodsCytology slides and cytopathology reports of 120 NSCLC patients who had undergone EBUS‐TBNA and lobectomy in our hospital system between 2015 and 2021 were retrospectively reviewed.ResultsOf 290 lymph nodes (110 cases) able to be reviewed, interpretation of 48 lymph nodes was discordant with the original cytopathology report. This included 31 lymph nodes originally reported as adequate, which were found to be non‐diagnostic on review. The diagnostic accuracy (benign/malignant) of lymph nodes that were sampled at EBUS‐TBNA and excised at surgery was 89%. Specific examination of cases where EBUS‐TBNA cytology did not reflect post‐surgical findings illustrated important features and limitations of the procedure. These included potential misclassification of lymph node stations, the presence of multiple, variably involved nodes at lymph node stations, and the failure to detect small volume disease.ConclusionsContinuous evaluation of EBUS‐TBNA performance identifies technical limitations and areas of improvement for cytopathology reporting. This is increasingly important in an era where lung cancer screening is expected to increase diagnosis of early‐stage disease and with the advent of novel treatments, including non‐surgical management options.