Objectives: (a) To assess whether differences in lung morphology observed in patients with adult respiratory distress syndrome (ARDS) are associated with differences in cardiorespiratory parameters, lung mechanics, and outcome. (b) To propose a new ARDS Severity Score to identify patients with a high mortality risk. Design: Prospective study over a 53-month period. Setting: Fourteen-bed surgical intensive care unit of a university hospital. Patients and participants: Seventy-one consecutive patients with early ARDS. Measurements ann results: Cardiorespiratory parameters were measured using a Swan-Ganz catheter, the pressure-volume (PV) curve was measured using the gross syringe method, and fast spiral computed tomography (CT) was performed. Patients with diffuse attenuations (n = 16) differed from patients with lobar attenuations (n = 26) regarding: (a) mortality rate (75 % vs. 42 %, p = 0.05), (b) incidence of primary ARDS (82 % vs. 50 %, p = 0.03), (c) respiratory compliance (47 +/- 12 vs. 64 +/- 16 mi per cmH(2)O(-1) p = 0.04), and (d) lower inflexion point (8.4 +/- 2.0 vs. 4.6 +/- 2.0 cmH(2)O, p = 0.001). A third group of patients with patchy attenuations (n = 29) had a mortality rate of 41 %, a respiratory compliance of 56 +/- 18 ml per cmH(2)O(-1) and a lower inflexion point of 6.3 +/- 2.7 cmH(2)O. The bedside chest radiograph accurately assessed lung morphology in only 42 % of the patients. In contrast to the scores based on the bedside chest radiograph, a new ARDS Severity Score based on CT lung morphology and cardiorespiratory parameters identified a subgroup of patients with a high mortality rate (greater than or equal to 60 %). Conclusions: In patients with ARDS, differences in lung morphology are associated with differences in outcome and lung mechanics. A new ARDS Severity Score based on CT lung morphology and cardiorespiratory parameters accurately identified patients with the most severe forms of ARDS and a mortality rate above 60 %.