A number of concepts in the CT evaluation of the thorax have been reviewed. The concepts have been presented through anatomic images generated on a specific CT system, but any equivalent system could have produced similar images. The important conclusions derived as a result of this review are: 1. Slice thickness, volume averaging and volume sampling are interrelated. Slice thickness and the orientation and inherent subject contrast of the anatomy to be studied usually determine the quality of the image obtained. 2. Thin section imaging (i.e., 5 mm and especially 1.5 mm) can demonstrate thoracic anatomy rarely (if ever) observed on thicker images (i.e., 10 mm). Although impractical for general analysis, thin section imaging provides insight into the potential limitations of a standard approach to thoracic evaluation. 3. CT analysis of the normal anatomy of the pericardium is incomplete. Although pathology related to the pericardium has been presented in the literature, more work is required to define: 1, the relationship, on CT, of the pericardium to structures within its confines; 2, abnormalities within the mediastinum that may affect the pericardium, and 3, lesions in the lung that may invade the mediastinum and pericardium. 4. The excellent contrast sensitivity of CT (aided by iodine infusion or bolus injection techniques and dynamic scanning) does not suffice to resolve certain problems in thoracic analysis. Extrapleural signs, the concept of invasion versus abutment, and the discrimination of benign from malignant mediastinal nodes continue to present diagnostic dilemmas in CT analysis of the thorax. 5. Interesting cases provide models for demonstrating CT pitfalls, mimics, and rarities. CT can be a valuable tool in resolving diagnostic dilemmas in some cases but creates dilemmas in others. 6. CT reformation images of the thorax can be dynamically generated on the scanner console. This hands-on technique can be a valuable tool for teaching anatomy in resident training programs.