1. A high-frequency high-flow ventilator has been developed which will produce abrupt changes in alveolar gas tensions. We have used it to study the individual contributions of PCO2 and PO2 in producing the oscillations which occur in the discharge of carotid chemoreceptors in the cat with respiration, by producing repeated end-tidal alternations (i) of PCO2 in constant hypoxia, (ii) of PO2 in constant normocapnia and (iii) of both PO2 and PCO2, i.e. of asphyxia. 2. The chemoreceptor response to alternations of PCO2 was always brisker than that to alternations of PO2 at 2, 4 or 8 s cycle durations. 3. An increase in the frequency of the alveolar alternation shortened the difference between the response times to PCO2 and PO2 but it increased the phase difference between the stimulus and the response waveforms. 4. With 4 s cycles, in normocapnic hypoxia, PCO2 was 2.9 times more effective (impulses s-1 Torr-1) than PO2 in producing oscillations in discharge. 5. The oscillations in discharge to simultaneous alternations of PO2 and PCO2 were not significantly different from the sum of individual oscillations to alternations of PCO2 and of PO2 alone. This was true with respect to timing and to amplitude of the oscillation. 6. Usually the amplitude of the chemoreceptor discharge oscillation in response to an asphyxial alternation was greater than the amplitude of the oscillation to either its PCO2 or its PO2 component alone. However, at the highest frequencies used, the phase relation between the PCO2 and PO2 components of the response could lead to the summed asphyxial response being less than its individual components. 7. The amplitudes and shapes of the oscillations in response to 4 s PCO2 alternations were not affected by changing either the steady-background PO2 or PCO2, but the amplitudes of the oscillations to pure PO2 alternations were enhanced by hypoxia and by hypercapnia. The importance of PO2 and PCO2 in giving rise to the natural respiratory oscillations in chemoreceptor discharge depends on the mean levels of the two gases. In normocapnic hypoxia (PO2 ca. 50 Torr) they are equally important but when PO2 is raised it becomes less important.