To investigate whether chest-wall mechanics could affect the total respiratory system pressure-volume (P-V) curve in patients with acute respiratory failure (ARF), and particularly the lower inflection point (LIP) of the curve, we drew the total respiratory system, lung, and chest-wall P-V curves (P-Vrs, P-VL, and P-VW, respectively) for 13 patients with ARF, using the supersyringe method together with the esophageal balloon technique. Measurements were randomly repeated at four different levels of positive end-expiratory pressure (PEEP) (0, 5, 10, 15 cm H2O) and from each P-V curve we derived starting compliance (Cstart), inflation compliance (Cinf), and end compliance (Cend). With PEEP of 0 cm H2O (ZEEP), an LIP on the P-Vrs curve was observed in all patients (7.5 +/- 3.9 cm H2O); in two patients an LIP was detected only on the P-VL curve (8.6 and 8.7 cm H2O, respectively); whereas in seven patients an LIP was observed only on the P-VW curve (3.4 +/- 1.1 cm H2O). In four patients, an LIP was detected on both the P-VL and P-VW curves (8.5 +/- 3.4 and 2.2 +/- 1.0 cm H2O, respectively). The LIP was abolished by PEEP, suggesting that a volume-related mechanism was responsible for the observed LIP on both the P-VL and P-VW curves. At high levels of PEEP, an upper inflection point (UIP) appeared on the P-Vrs and P-VL curves (11.7 +/- 4.9 cm H2O and 8.9 +/- 4.2 cm H2O above PEEP, respectively) suggesting alveolar overdistension. In general, PaO2 increased with PEEP (from 81.7 +/- 35.5 mm Hg on ZEEP to 120 +/- 43.8 mm Hg on PEEP 15 cm H2O, p < 0.002); however, the increase in PaO2 with PEEP was significant only in patients with an LIP on the P-VL curve (from 70.5 +/- 16.2 mm Hg to 117.5 +/- 50.7 mm Hg, p < 0.002), the changes in PaO2 in patients without an LIP on the P-VL curve not being significant (from 91.3 +/- 45.4 mm Hg to 122.2 +/- 41.1 mm Hg). We conclude that in ventilator-dependent patients with ARF: (1) the chest-wall mechanics can contribute to the LIP observed on the P-Vrs curve; (2) the improvement in PaO2 with PEEP is significant only in patients in whom LIP is on the lung P-V curve and not on the chest wall curve; (3) high levels of PEEP may overdistend the lung, as reflected by the appearance of a UIP; (4) measurement of P-Vrs alone may be misleading as a guide for setting the level of PEEP in some mechanically ventilated patients, at least in the supine position, although it helps to prevent excessive alveolar overdistension by indicating the inflection volume above which UIP may appear.