Tracheal gas insufflation (TGI) improves the efficiency of CO2 elimination accomplished by conventional mechanical ventilation, primarily by reducing the anatomic (series) dead space volume. Dead space proximal to the catheter tip can be reduced by two methods. Fresh gas introduced at the carinal level during inspiration may effectively "bypass" the upper airway. Alternatively, proximal dead space can be "washed out" with fresh gas during expiration to reduce CO2 rebreathing. We examined these two modes of TGI-aided dead space reduction in nine paralyzed normal dogs receiving conventional mechanical ventilation and compared these results to those obtained with a catheter that delivered fresh gas continuously at the same flow rate, thereby accomplishing both bypass and washout. Total inspired tidal volume and cycling frequency were held constant. Differences in CO2 elimination efficiency among the TGI modes were flow dependent. Continuous catheter flow at 5 or 10 L/min reduced PaCO2 and physiologic dead space fraction (VD/VT) more than either proximal bypass or end-expiratory washout (p < 0.001). At the same catheter flow settings expiratory washout tended to improve VD/VT more than did inspiratory bypass. Under the conditions tested, constant tracheal insufflation of fresh gas improves alveolar ventilation by mechanisms that include, but are not limited to, a functional reduction in the dead space proximal to the catheter tip.