Transbronchial needle aspiration has emerged as a key technique for sampling mediastinal adenopathy but variable yields are reported. To determine the number of aspirates needed to optimize yield, we prospectively studied transbronchial needle aspiration and the sequential effect of each successive specimen on diagnostic yield in 79 patients with known or suspected lung carcinoma and mediastinal adenopathy. A total of 451 aspirates were performed in 79 patients (mean, 5.7 aspirates per patient; range, 2-13) with 45 cases (57%) positive for malignancy. A cytologically positive transbronchial needle aspiration occurred with the first aspirate in 42% of patients in whom this procedure established mediastinal nodal involvement. All positive results were achieved with seven or fewer aspirates. Similar yields were obtained for small cell and non-small cell lung cancer after seven aspirates. Rapid on-site specimen cytologic evaluation was used in 55 of 79 cases (70%), with a positive diagnosis obtained in 39 of 55 cases (71%) with on-site evaluation compared with six of 24 cases (25%) performed without on-site evaluation. The data suggest there is a plateau in yield after seven transbronchial needle aspirates, which may be sufficient to obtain an optimal yield in assessing patients with lung cancer and mediastinal adenopathy.