In the prone position, computed tomography scan densities redistribute from dorsal to ventral as the dorsal region tends to reexpand while the ventral zone tends to collapse. Although gravitational influence is similar in both positions, dorsal recruitment usually prevails over ventral derecruitment, because of the need for the lung and its confining chest wall to conform to the same volume. The final result of proning is that the overall lung inflation is more homogeneous from dorsal to ventral than in the supine position, with more homogeneously distributed stress and strain. As the distribution of perfusion remains nearly constant in both postures, proning usually improves oxygenation. Animal experiments clearly show that prone positioning delays or prevents ventilation-induced lung injury, likely due in large part to more homogeneously distributed stress and strain. Over the last 15 years, five major trials have been conducted to compare the prone and supine positions in acute respiratory distress syndrome, regarding survival advantage. The sequence of trials enrolled patients who were progressively more hypoxemic; exposure to the prone position was extended from 8 to 17 hours/day, and lung-protective ventilation was more rigorously applied. Single-patient and meta-analyses drawing from the four major trials showed significant survival benefit in patients with PaO2/FiO2 lower than 100. The latest PROSEVA (Proning Severe ARDS Patients) trial confirmed these benefits in a formal randomized study. The bulk of data indicates that in severe acute respiratory distress syndrome, carefully performed prone positioning offers an absolute survival advantage of 10-17%, making this intervention highly recommended in this specific population subset.