Computed tomography (CT) has been utilized to study acute respiratory distress syndrome (ARDS) since the middle 1980s, when it revealed the inhomogeneous pattern of the lung lesion. Its advantages rely on the strict correlation between CT density and the lung physical density, allowing a quantification of lung compartments with different degrees of aeration. By CT scans, ARDS lung appeared to be “small” rather than “stiff,” leading to the “baby lung” concept. The regional analysis revealed that this appearance derives from an evenly distributed lung edema, which tends, because of gravitational forces, to lie predominantly in the most dependent regions, leading to alveolar collapse. New data suggest that such a “sponge lung” is made by a “core,” consolidated, lung portion, from which, through an inflammatory reaction, lung edema will spread, determining the collapsed and recruitable lung portion. The amount of recruitable lung varies among ARDS patients. This knowledge is necessary for a rational positive end-expiratory pressure (PEEP) setting because the amount of tissue maintained aerated by PEEP is closely associated with the amount of recruitable lung. CT scans may also help to diagnose ARDS because CT provides a good estimate of the high-permeability lung edema, the characteristic lesion of this syndrome.