Measurement of mean lung volume (MLV) in high-frequency oscillatory ventilation (HFO) may be useful for optimizing the high lung volume strategy, but has not been available until now. We have measured MLV by means of the sulfur hexafluoride (SF6) washout method in 13 premature infants ventilated with HFO because of respiratory distress syndrome (gestational age, 23 to 31 wk [median, 25 + 6 / 7 wk]; birthweight, 630 to 1,140 g [790 g]; age at measurement, 2 to 10 d [4 d]; weight, 675 to 1,250 g [850 g]). To evaluate the relationship between MLV and mean airway pressure (MAP), the latter was systematically varied between the measurements. With clinically selected MAP, MLV was between 23.3 and 41.9 ml / kg (median, 33.5 ml / kg) and was strongly dependent on MAP in each patient; linear regression analyses resulted in slope factors between 1.0 and 6.9 ml / cm H2O / kg (median, 2.83 ml / cm H2O / kg), with correlation coefficients between 0.77 and 0.99 (median, 0.94). Stabilization of MLV after modification of MAP took 2 to 25 min (median, 9 min). We conclude that the selection of MAP on a clinical basis leads to a wide range of different MLVs. Despite the strong dependence of MLV on MAP, the prediction of MLV solely based on MAP was impossible because of large patient to patient variability of linear regressions. Alveolar recruitment and derecruitment may take up to 25 min after MAP changes.