To investigate the relationship between lung anatomy and pulmonary mechanics in acute respiratory failure (ARF), 20 patients with ARF underwent computerized tomography (CT) at 3 levels of positive end-expiratory pressure (PEEP) (5, 10, and 15 cm H2O). The static pressure-volume curve of the total respiratory system and the lung volumes (helium dilution method) were also measured. By knowing the lung volumes and analyzing the CT number frequency distribution, a quantitative estimate of normally aerated, poorly aerated, and nonaerated lung tissue was obtained at each level of PEEP. The recruitment was defined as the percent increase of normally aerated tissue from 5 to 15 cm H2O. We found that the different compliances (starting compliance, inflation compliance, and deflation compliance) were correlated only with the amount of normally aerated tissue present in the range of pressures explored by a given compliance (5 cm H2O for starting compliance and 15 cm H2O for inflation and deflation compliances). No relationship was found between the compliances and the poorly aerated and nonaerated tissue. The specific compliance was in the normal range, whereas the amount of recruitment was related to the ratio of inflation compliance to starting compliance. Our data suggest that (1) the pressure-volume curve parameters in ARF investigate only the residual healthy zones of the lung and do not directly estimate the "amount" of disease (poorly or nonaerated tissue), (2) the pressure-volume curve may allow an estimate of the anatomic recruitment, and (3) the residual normally aerated zones of the ARF lung seem to maintain a normal intrinsic elasticity.