To evaluate the individual and additive diagnostic yield(s) of several bronchoscopic sampling techniques for the diagnosis of lung lesions with no corresponding airway abnormalities, consecutive patients with lung nodules or masses were prospectively evaluated between December 1989 and November 1994. A CT of the chest was done in all patients before flexible bronchoscopy (FB). Size, location, and character of the border of the lesion were determined. During FB, using biplane fluoroscopic guidance, the lesion was localized and following sampling techniques were done: brushing, transbronchial lung biopsy (TBLB), and Sofcor transbronchial needle aspiration (STBNA). Bronchial washings (BWs) were collected throughout the procedure. Problems associated with each sampling technique were noted. Forty-nine patients underwent 51 FB. A diagnosis was established by FB in 36 (73%). After a nondiagnostic FB, histologic diagnosis was established in 9 of 13 patients by other methods. A benign or malignant nature of lesion was established in other four patients by clinical follow-up. FB was diagnostic in 32 of 40 (80%) patients with primary lung cancer, in 3 of 6 (50%) patients with benign disease, and in 1 of 3 (33%) patients with metastatic disease to lung. All sampling procedures could be done in 33 of 51 (65%) FBs. Overall diagnostic yields were as follows: BW, 18 of 51 (35%); brush, 25 of 48 (52%); TBLB, 23 of 40 (57%); and STBNA, 19 of 37 (51%). In 12 of 51 (24%) FBs, only one sample was diagnostic. Lesions with sharp borders had a lower combined diagnostic yield, 13 of 24 (54%) compared to lesions with fuzzy borders, 20 of 24 (83%) (p=0.03). Yield of TBLB in lesions with fuzzy borders, 14 of 18 (78%), was higher than the yield for lesions with sharp borders, 6 of 19 (32%) (p=0.005). Size of the lesion in centimeters in patients with a positive FB (4.55+/- 2.35; mean +/-SD) was significantly larger than in patients with a nondiagnostic FB (3.14+/-1.31; mean+/-SD) (p=0.009, t test). Diagnostic yield was directly related to the size of the lesion. For lesions less than 2 cm, yield was 6 of 11 (54%) (p=0.19); for lesions less than 3 cm, yield was 12 of 21 (57%) (p=0.07); and for lesions greater than 3 cm, yield was 24 of 30 (80%). Yield from lesions located in the lower lobe basal segments or the apical segment of upper lobes was lower (11/19, 58%) than that from lesions in other parts of the lung (25/30, 83%) (p=0.05). FB was terminated prior to collecting all samples because of severe bleeding after brushing (n=3) or instability of the patient (n=4). None of the patients required intubation. There were no pneumothoraces. Diagnostic yield of FB depends on the location, size, character of the border of the lesion, and the ability to perform all sampling methods. Brushing, TBLB, and STBNA should be performed in all patients to give the best diagnostic yield. Routine cytologic examination of BW is unnecessary. Methods other than FB should be considered for lesions 2 cm or less in size, especially when they have a sharp border and/or are located in the basal segment of a lower lobe/apical segment of an upper lobe.