We reviewed the ventilatory responses obtained from rebreathing experiments on a population of 22 subjects. Our aim was to derive parameter estimates for an 'average subject' so as to model the respiratory chemoreflex control system. The rebreathing technique used was modified to include a prior hyperventilation, so that rebreathing started at a hypocapnic P(CO2) and ended at a hypercapnic P(CO2). In addition, oxygen was added to the rebreathing bag in a controlled manner to maintain iso-oxia during rebreathing, which allowed determination of the response at several iso-oxic P(O2) levels. The breath-by-breath responses were analysed in terms of tidal volume, breathing frequency and ventilation. As P(CO2) rose, ventilation was first steady at a basal value, then increased as P(CO2) exceeded a breakpoint. We interpreted this first breakpoint as the threshold of the combined central and peripheral chemoreflex responses. Above, ventilation increased linearly with P(CO2), with tidal volume usually contributing more than frequency to the increase. When breathing was driven strongly, such as in hypoxia, a second breakpoint P(CO2) was often observed. Beyond the second breakpoint, ventilation continued to increase linearly with P(CO2) at a different slope, with frequency usually contributing more than tidal volume to the increase. We defined the parameters of the variation of tidal volume, frequency and ventilation with P(O2) and P(CO2) for an average subject based on a three-segment linear fit of the individual responses. These were incorporated into a model of the respiratory chemoreflex control system based on the general scheme of the 'Oxford' model. However, instead of considering ventilatory responses alone, the model also incorporates tidal volume and frequency responses.