PURPOSE: The purpose of this study was twofold. In a preliminary study, we evaluated the accuracy of 3-D(three-dimensional) CT for the estimation of pneumothorax volume and for providing the optimal postprocessing method for clinical study. In the clinical study, we determined which of the two methods, Collins' and Axel's, was more accurate for the estimation of pneumothorax volume, as seen on chest radiographs, using 3-D CT as the standard of reference. MATERIALS AND METHODS: In the preliminary study, 3-D CT was applied to phantoms and to four patients with p-neumothorax using two different postprocessing methods, manual contour delineation and thresholding. In the clinical study, 3-D CT was performed in 13 patients with pneumothorax. For the purpose of evaluating conventional radiographs, a localizer scan was used for comparing the accuracy of Collins' method with that of Axel's method, with 3-D CT as the standard of reference. RESULTS: The preliminary study revealed that 3-D CT estimated pneumothorax volume with great accuracy and that manual contour delineation and thresholding measured volume equally well. Because of the shorter postprocessing time required with thresholding than with manual contour delineation (5 min versus 30 min), the former was used during clinical study. The results of this indicated close correlation between the measurements obtained using Collins' method on chest radiographs and those obtained by 3-D CT(r=0.95, p<0.05). In contrast, measurements obtained using Axel's method correlated poorly with those obtained by 3-D CT (r=0.55, p>0.05). CONCLUSION: 3-D CT can estimate pneumothorax volume with great accuracy. Collins' method is superior to Axel's method for the quantification of pneumothorax volume as seen on chest radiographs.