Computed tomography (CT) is an obvious modality for subclassification of COPD. Traditionally, the pulmonary involvement of chronic obstructive pulmonary disease (COPD) in smokers is understood as a combination of deleterious effects of smoking on small airways (chronic bronchitis and small airways disease) and distal to the airways with destruction and loss of lung parenchyma (emphysema). However, segmentation of airways is still experimental; with contemporary high-resolution CT (HRCT) we can just see the "entrance" of small airways, and until now changes in airway morphology that have been observed in COPD are subtle. Furthermore, recent results indicate that emphysema may also be the essential pathophysiologic mechanism behind the airflow limitation of COPD. The definition of COPD excludes bronchiectasis as a symptomatic subtype of COPD, and CT findings in chronic bronchitis and exacerbations of COPD are rather unspecific. This leaves emphysema as the most obvious candidate for subclassification of COPD. Both chest radiologists and pulmonary physicians are quite familiar with the appearance of various patterns of emphysema on HRCT, such as centrilobular, panlobular, and paraseptal emphysema. However, it has not yet been possible to develop operational definitions of these patterns that can be used by computer software to automatically classify CT scans into distinct patterns. In conclusion, even though various emphysema patterns can be recognized visually, CT has not yet demonstrated a great potential for automated subclassification of COPD, and it is an open question whether it will ever be possible to achieve success equivalent to that obtained by HRCT in the area of interstitial lung diseases.