OBJECTIVE Static or quasi-static pressure-volume (P-V ) curves can be used to determine the lung mechanical properties of patients suffering from acute respiratory distress syndrome (ARDS). According to the traditional interpretation, lung recruitment occurs mainly below the lower point of maximum curvature (LPMC) of the inflation P-V curve. Although some studies have questioned this assumption, setting of positive end-expiratory pressure 2 cmH2O above the LPMC was part of a 'lung-protective' ventilation strategy successfully applied in several clinical trials. The aim of our study was to quantify the amount of unrecruited lung at different clinically relevant points of the P-V curve. METHODS P-V curves and electrical impedance tomography (EIT) data from 30 ARDS patients were analysed. We determined the regional opening pressures for every EIT image pixel and fitted the global P-V curves to five sigmoid model equations to determine the LPMC, inflection point (IP) and upper point of maximal curvature (UPMC). Points of maximal curvature and IP were compared between the models by one-way analysis of variance (ANOVA). The percentages of lung pixels remaining closed ('unrecruited lung') at LPMC, IP and UPMC were calculated from the number of lung pixels exhibiting regional opening pressures higher than LPMC, IP and UPMC and were also compared by one-way ANOVA. RESULTS As results, we found a high variability of LPMC values among the models, a smaller variability of IP and UPMC values. We found a high percentage of unrecruited lung at LPMC, a small percentage of unrecruited lung at IP and no unrecruited lung at UPMC. CONCLUSIONS Our results confirm the notion of ongoing lung recruitment at pressure levels above LPMC for all investigated model equations and highlight the importance of a regional assessment of lung recruitment in patients with ARDS.