The ventilatory responses to steady-state venous CO2 loading (iv CO2) and CO2 inhalation have been observed in chloralose-urethan-anesthetized dogs. Intravenous CO2 was administered by increasing the CO2 fraction of gas ventilating a membrane gas exchanger in an arteriovenous bypass; blood flow rate was fixed at 30 ml/min. During the study, we identified a time-dependent hyperventilation in all 14 experimentally treated dogs and in 4 additional sham-treated dogs. When we tested 8 of these animals with a protocol having small progressive increments in iv CO2 loading rate, we observed a response approaching isocapnia during iv CO2 and a large hypocapnia when we returned to control conditions. The use of a randomized protocol in 6 animals demonstrated the necessity of accounting for this systematic base-line shift, because before doing so the response depended more on the passage of time than on the nature of the CO2 load. After this analytical adjustment was made, there was no significant difference between the respiratory controller gains (delta nu E/delta Paco2) for inhaled and iv CO2.