A correlative review was made of the type of cytology specimens (sputum, bronchial washing and bronchial brushing) together with the corresponding histopathologic specimens of 108 patients. One hundred patients had primary pulmonary carcinomas diagnosed histopathologically (84) or clinically (16); 5 had carcinomas metastatic to the lungs and 3 had apparently false-positive cytologic results for lung cancer. The correlative review was used to determine the diagnostic reliability of pulmonary cytopathologic techniques in the detection and classification of lung carcinomas (i.e., the sensitivity and accuracy). The overall sensitivities of sputum, bronchial washing and bronchial brushing cytology were 60%, 66% and 77%, respectively (p less than 0.05). Bronchial brushing had a higher sensitivity (80%) for peripheral and metastatic lesions than did sputum (37%) or bronchial washing (60%). The overall accuracies of sputum, bronchial washing and bronchial brushing cytology were 79%, 75% and 76%, respectively, which is not statistically different. Regardless of the sampling methods, cytologic typing of squamous-cell and small-cell carcinomas was highly accurate but was less satisfactory for the other types of lung carcinomas. In the 16 cases in which endoscopic biopsies were either not attempted or gave negative results, one or more pulmonary cytologic specimens showed malignant cells. It is concluded that: (1) pulmonary cytopathologic techniques have excellent sensitivity and accuracy in the diagnosis of lung carcinomas; (2) they may establish the diagnosis of pulmonary carcinomas when endoscopic biopsies give negative results; and (3) they are particularly helpful in cases in which endoscopic biopsies suffer from a low yield (peripheral lesions) or create a considerable danger to the patients (iatrogenic hemorrhage).