Deciding when to wean neonates from extracorporal membrane oxygenation (ECMO) can be difficult. The usefulness of simple measurements of pulmonary mechanics e.g., dynamic compliance (Cdyn) has been questioned. We investigated the pulmonary mechanics of eight neonates using the interrupter technique, which allows the partitioning of pulmonary mechanics into compartments representing the conducting airways and more peripheral phenomena (viscoelastic properties and "pendelluft"). Three neonates required ECMO for a congenital diaphragmatic hernia (CDH), two for hyaline membrane disease (HMO), two for meconium aspiration syndrome (MAS), and one for pneumonia. All neonates with MAS, HMD, and pneumonia were successfully weaned from ECMO when their Cdyn was 0.3 mL/cmH2O/kg or greater [mean 0.34 +/- 0.06 (SEM)]. All three neonates with CDH died and their highest Cdyn was 0.21, 0.19, and 0.09 mL/cmH2O/kg respectively (mean, 0.16 +/- 0.037). The airway resistance (Raw) and the slower component of pressure change after interruption (delta Pdiff), a measure of the more peripheral phenomena of the lung, were not significantly different in those neonates who survived and those who did not. The values for delta Pdiff in all patients were higher than those in healthy neonates. However, the Raw was not different. This suggests that the major disturbance in pulmonary mechanics was distal to the conducting airways. Those neonates who were successfully weaned from ECMO had a significantly higher Cdyn 24-48 hours prior to decannulation. Considering the lung as a two-compartment model offers no advantages when compared to the one-compartment model for the prediction of the outcome of a neonate on ECMO.