The traditional practice of using high inflation pressures to maintain normal tidal volumes and arterial blood gases has been encouraged by the perception of uniformly distributed damage in acute lung injury. Although the frontal chest radiograph often suggests uniformity, recent work highlights the heterogeneous pathoanatomy and lung mechanics that actually characterize the adult (acute) respiratory distress syndrome. This heterogeneity is important to consider when applying mechanical ventilation, because impressive experimental evidence strongly indicates the potential for traditional selections for volume and pressure to impede lung healing or extend damage to previously unaffected areas. Because lung regions differ markedly with regard to distensibility and fragility, the acutely injured lung should be viewed as small rather than stiff. Aerated lung units appear to have nearly normal gas-to-tissue ratios and well-preserved mechanical and gas-exchanging properties. Mechanical ventilation may expose endothelial and epithelial barriers to excessive stress, allowing proteinaceous alveolar edema to form without actual membrane rupture. Such damage has been linked experimentally to an excessive transalveolar inflation pressure, to a tidal volume inappropriate to the size of the aeratable lung mass, or to damaging shear forces that develop when insufficient end-expiratory alveolar pressure is maintained to prevent tidal opening and reclosure of susceptible alveoli. Pathologic, physiologic, and theoretical arguments favor a strategy that attempts to avoid tidal alveolar collapse and to keep transalveolar pressure (not PaCO2) within normal physiologic limits. CO2 retention may be an unavoidable consequence of such a lung-protection strategy. Although the traditional paradigm for ventilation appears in need of revision, it must be recognized that few prospective, controlled trials of alternative ventilation modes have been undertaken to prove their superiority.