Background : We assessed the accuracy of staging in evaluation of bronchial invasion, thus found the role of CT in patients who underwent resective surgery in primary lung cancer. Materials and Methods : Authors retrospectively analized the preoperative CT scans of 156 patients receiving pneumonectomy(n = 95) and lobectomy(n = 61). Among lobectomy patients, 7 patients subsequently performed pneumonectomy because of positive resection margin of bronchus in frozen biopsy. We also retrospectively analized CT scans of non-operated 60 patients who performed sufficient bronchoscopic biopsy. Bronchial wall thickness more than 3mm, irregular wall thickening and reduction of diameter by CT were defined as bronchial invasion. The pathologic examination of resection margin were positive in 20, stump recurrence occurred in 6 of the operated group, and the pathologic examination of biopsy of bronchial wall were positive in 34 of the non operated group, and these were an regarded as bronchial invasion. Results : The CT assessment of bronchial invasion revealed low sensitivity (11.5%), low positive predictability(38%), but high specificity(96%) and relatively high accuracy (84%) in the operated group and higher sensitivity (62%), higher positive predictability(95%) in non-operated group. Conclusion : In lung cancer patients who underwent operation, CT showed very low sensitivity and positive predictability in evaluation of bronchial invasion. Because the usefulness of CT in evaluation of bronchial invasion is limited, therefore aggressive fiberoptic bronchoscopic biopsy is thought to be necessary before surgical attempt.