BACKGROUND Conventional endoscopic transbronchial needle aspiration (TBNA) is a common procedure used to obtain samples for diagnosing and staging lung lesions. Recently, endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) has been developed and increasingly used by clinicians. Clinical data suggest that EBUS-TBNA has higher sensitivity and specificity than conventional TBNA in staging lung cancers. In this study, the authors have investigated the cytological features and compared the diagnostic yield of these procedures in lung cancer patients. METHODS A computer search of the cytopathology archives at The Johns Hopkins Hospital revealed 188 EBUS-TBNA cases (308 lymph nodes; 47 lung lesions) and 74 TBNA cases (106 lymph nodes; 44 lung lesions) over a 28-month period. All cytological material was correlated with available corresponding surgical material. RESULTS The most frequently sampled lymph nodes were stations 4R and 7 in both TBNA and EBUS-TBNA; in addition, the EBUS-TBNA showed a wide range of lymph node sampling. EBUS-TBNA had a significantly lower nondiagnostic rate (8.7%) compared with TBNA (28.3%, P < .05) in staging lung cancers. In lymph node sampling, the sensitivity and specificity were 54.5% and 100% in the TBNA group and 85.2% and 100% in the EBUS-TBNA group. In lung specimens, the nondiagnostic rates of TBNA and EBUS-TBNA were 6.8% and 4.3%, respectively (P > .05), and the sensitivity and specificity were 78.9% and 100% in the TBNA group; and 89.5% and 100% in the EBUS-TBNA group. CONCLUSIONS Findings indicated that EBUS-FNA cytology is an optimal modality for diagnosing and staging in lung cancer patients, in comparison to conventional TBNA.