PURPOSE: To evaluate the role of CT and CT features in the diagnosis of bronchogenic carcinomas not detected by plain radiography. MATERIALS AND METHODS: Eighteen patients [19 primary cancer lesions, M:F=16:2, aged 43 -75 (mean, 56.3)years] with lung cancer initially not detected by plain radiography were involved in this study. CT scanning was performed in all cases, and fibrobronchoscopy, and sputum cytology. each in 17. Lesions were divided into two groups: the central type, if on or proximal to the segmental bronchus, and the peripheral type, if distal to this. Plain radiographs were analysed for possible causes of occultness and for clinical characteristics including cell type, location, and size. We focused on the CT findings, comparing cases undetected by CT with those undetec6ted by bronchoscopy. RESULT: In the central type, the cause of occultness, as seen on plain radiographs, was small size, no secondary findings, or confusing shadow from hilar vessels. In the peripheral type, the cause was overlapping shadow due to normal structures of the chest, or combined diseases. Eight lesions were first detected by sputum cytology, 6 by bronchoscopy, and 5 by CT. Fourteen lesions were the central type (main bronchus 2, lobar bronchus 7, segmental bronchus 5), and five were peripheral. Central-type lesions were either squamous cell carcinoma (n =11), adenocarcinoma (n =1), small cell carcinoma (n =1), or large cell carcinoma (n =1). The peripheral type were either squamous cell carcinoma (n =2), adenocarcinoma (n =2), or large cell carcinoma (n =1). Size ranged from 0.2 to 4(mean, 2; central 1.7, peripheral 2.8) cm. Surgical resection was possible in 15 patients (16 cancers, including 13 at stage I). Only two were at a stage which rendered them unresectable. CT revealed 13 cancers, including all those which were peripheral. The findings were endobronchial nodule (n =4), bronchial wall thickening (n =1), perihilar mass (n =3), parenchymal mass (n =2), and subpleural mass (n =3). In six central-type cases [endobronchial mass (n =5), carcinoma in situ(n =1)], CT revealed no evidence of cancer. The mean size of these lesions was 1.1cm, and all were stage I. Bronchoscopy failed to detect five cases, including four peripheral cancers and one central. The mean size of these was 2.7 cm and all three adenocarcinomas were included in this group. In two of the five cases in which sputum cytology showed negative results, the existing condition was revealed by CT. CONCLUSION: For the detection of peripheral lung cancer, CT is better than bronchoscopy, though in cases of central lung cancer, in which CT plays a complementary role, bronchoscopy is better than complementary to bronchoscopy which is more excellent than CT in detecting central lung CT. In 68% of cases, CT revealed lung cancer which was not detected by plain radiography, and is therefore a suitable noninvasive screening method for the detection of this cancer.