The objective of this study was to evaluate carbon dioxide (CO2) clearance, delivered pressures and tidal volume (VT) during neonatal nasal intermittent positive pressure ventilation (NIPPV) with two commonly used interfaces.A neonatal lung model, with either short binasal prongs (SBP) or a small caliber nasal cannula (RAM) interface, was tested over a range of clinically relevant settings. A fixed amount of CO2 was infused and the fraction remaining in the lung 100 s postinfusion was measured. Pressure transmission to the lung and VT was measured at the level of the trachea.CO2 elimination was directly proportional to the inspiratory pressure during NIPPV. At peak pressures of 22 to 34 cm H2O, CO2 clearance was greater (P<0.001) with SBP as compared with RAM. Relative to the set ventilator parameters, a substantial pressure dampening effect was documented at the lung level, which was significantly lower with RAM when compared with SBP (2.8% (0.2) versus 11.9% (1.5), P<0.0001). CO2 elimination was dependent on VT and effective despite only a small fraction of physiological VT (maximum delivered VT%: SBP 15.5 (0.7) versus RAM 6.1 (1.4), P<0.0001).NIPPV promotes CO2 elimination even at low transmitted airway pressures, but less effective with RAM as compared with SBP. CO2 elimination despite small VT suggests that NIPPV may depend on a non-conventional gas-exchange mechanism.