The dose effect of caffeine (10-70 mg/kg iv) on pulmonary ventilation (VE), mean inspiratory flow (VT/TI), and tracheal pressure generated 0.3 and 0.5 s (P0.3 and P0.5, respectively) after the onset of inspiration against airway occluded at end expiration was studied in cats anesthetized with pentobarbital sodium (35 mg/kg ip) breathing various gas mixtures. With air and 50% O2 (balance N2), increasing doses of caffeine caused a progressive increase in VE that was associated with a reduction in end-tidal PCO2. When the latter was maintained at control (precaffeine) level by inhalation of CO2, the increase in VE was, at all caffeine levels, about three times that under nonisocapnic conditions. Both under isocapnic and nonisocapnic conditions the greatest incremental changes of VE were observed after administration of the first 10-mg/kg aliquot of caffeine, i.e., the current acceptable clinical dose. In all instances, the changes in VE were proportionally the same as the corresponding changes in VT/TI, P0.3, and P0.5, suggesting that caffeine did not appreciably alter either the shape of the inspiratory driving pressure waveform or the impedance of the respiratory system but simply acted by increasing the amplitude of the neuromuscular inspiratory output. An additive interaction between caffeine and end-tidal PCO2 was observed in the VE, VT/TI, and P0.3 responses at levels of CO2 at or below the eucapnic range.