In a retrospective analysis of 50 patients with segmental or lobar atelectasis of the lung, chest radiographs and CT studies were compared for their abilities to distinguish whether a centrally obstructing tumor was the cause. This was done to help define the role of CT in evaluating patients with atelectasis. Atelectasis was caused by an obstructing tumor in 27 cases and a variety of other conditions in 23. The chest radiograph correctly identified an obstructing tumor as the cause of atelectasis in 24 of 27 patients on the basis of the presence of a central hilar mass or obvious bronchial abnormality; there was 89% sensitivity and 96% specificity with a 12% false-negative rate and a 4% false-positive rate. CT correctly identified all 27 obstructing carcinomas on the basis of the presence of either a central bronchial abnormality or a central hilar mass; there was 100% sensitivity and 87% specificity with a 0% false-negative rate and a 10% false-positive rate. Absence of air bronchograms or the presence of mucus-filled bronchi within the atelectatic lung were secondary CT findings that also favored the presence of an obstructing tumor. Although the chest radiograph was more specific than CT for tumor as the cause of atelectasis (96% vs 87%, respectively), it was less sensitive than CT for tumor (89% vs 100%, respectively) resulting in missed tumor diagnoses. CT identified all cases caused by obstructing tumor and successfully excluded obstructing tumor in most of the remaining cases, with an acceptable number of false-positive tumor diagnoses (10%). CT should be performed when the cause of segmental or lobar atelectasis cannot be established with certainty on the basis of the chest radiograph.