A group of 168 consecutive lung cancer patients in whom a definitive diagnosis of primary lung cancer was established either in a conventional cytologic specimen of sputum or bronchial material or in a specimen obtained by fine needle aspiration (FNA) biopsy was reviewed to compare the relative accuracies between the modalities of sputum and bronchial material on one hand versus FNA cytology on the other in the diagnosis of lung cancer. The patients included in the study were selected from a total of 1,093 patients who had been diagnosed and treated for lung cancer at Duke University Medical Center over the five-year period of January 1, 1980, through December 31, 1984. In 325 (29.8%) of the 1,093 patients, a definitive cancer diagnosis was established from histopathologic study alone, without any cytologic diagnoses. In 420 patients (38.4%), both histologic and cytologic material had been interpreted as being conclusively diagnostic for lung cancer. In 348 patients (31.8%), a cytologic diagnosis of lung cancer was made without a histologic confirmation. Thus, in a total of 768 (70.3%) of the 1,093 cases, a definitive cytologic diagnosis of cancer had been made. Of these 768 patients, 168 had been evaluated by both conventional respiratory cytologic methods (examination of sputum and bronchial material) and with FNA biopsy cytology. In 9 patients (5.4%), only conventional respiratory cytologic specimens were conclusively diagnostic for cancer. In 122 patients (72.6%), only the FNA biopsy specimen was diagnostic. In 37 patients (22.0%), both conventional respiratory specimens and FNA specimens yielded a definitive lung cancer diagnosis. The FNA specimen was the only positive cytologic specimen in 90.2% of large cell undifferentiated carcinomas, 79.5% of adenocarcinomas, 66.7% of small cell undifferentiated carcinomas and 58.2% of squamous cell carcinomas. In 26.5% of the patients, a diagnosis of cancer could have been established on conventional cytologic specimens, without the necessity of proceeding to percutaneous FNA biopsy. From this study, it is concluded that the techniques of conventional respiratory cytology and FNA biopsy cytology are complementary in the diagnosis of lung cancer. While the percentage of lung cancers diagnosed by FNA biopsy cytology alone is much greater than that obtained by conventional respiratory cytology alone, more than one-fourth of these cancers could be detected by the less invasive techniques of sputum collection and bronchoscopy.