Ventilatory response to inhaled CO2 whose concentration varied sinusoidally was determined in a helathy male subject. The frequency, amplitude and mean level of inhaled CO2 were varied. Relative maxima were determined and a repetitive averaging technique was used to maximize the signal to noise ratio. Four to one hundred complete cycles (500 to 650 breaths) were studied in this manner, producing alveolar ventilation curves of about equal quality. Lissajous plots were made for each pair of the three variables (VA, CI, CA) and compared to similar plots obtained from a three compartment analog model of the respiratory control system that was also forced sinusoidally. The model as well as the experimental results revealed a third harmonic in the respiratory response-probably due to the nonlinearity of the system. The ratio of the amplitudes of the three variables was plotted for each possible pair vs the frequency. Thus, the magnitude portion of the describing function was plotted as a function of frequency for each of the three variables. These magnitude portions of the describing function were similar to those predicted by the model and indicated a breakpoint at about .05 rad/sec. This method will be helpful in obtaining precise data for studying the respiratory control system. © 1969.