Ventilatory responses to changes in PCO2 of the blood perfusing the central nervous system were studied breath by breath by pneumotachography in Pekin ducks under transient and steady condition. 1. Transients. In conscious birds, all the arteries to the cephalic region were tied or clamped, except the right internal carotid. The blood supply via the single remaining arterial pathway was transiently replaced, for about 15 sec, by injecting 2 ml of blood previously made either normocapnic (control PCO2 = 32 Torr) or hypercapnic (test; PCO2 = 76 Torr) from a syringe thermostated at 41 degrees C, under normal oxygenation (PO2 around 110 Torr) and mean endovascular pressure (107 mm Hg). During control injections, no significant ventilatory changes were observed. In contrast, test injections provoked an early and significant 20% increase in the minute volume of ventilation. 2. Steady conditions. Using cross-perfusion between pairs of anesthetized ducks, the head of a recipient animal (R) was vascularly isolated from the trunk and perfused by a donor (D), the nervous connections with the trunk remaining intact. When giving some CO2 to breathe to D (FICO2 = 0.05) while R breathed ambient air, arterial PCO2 increased in D and in the head of R, and hyperventilation occurred in both ducks. As a consequence of this hyperventilation, PCO2 decreased in the arterial blood and the end-tidal gas of R.