A technique is described which permits the inscription of the ventilatory response to isocapnic hypoxia in man as a continuous curve relating alveolar oxygen tension and minute ventilation. The adjustment of ventilation to changes in alveolar oxygen tension is complete in 18-23 sec and this is sufficiently rapid to justify the use of a non-steady-state method. Changes in alveolar carbon dioxide tension are prevented by addition of carbon dioxide to the inspired gas. The resulting [unk]V(E)-P(Ao2) curves are hyperbolic such that falling P(Ao2) produces only slight rises in [unk]V(E) until a critical P(Ao2) range of 50-60 mm Hg is reached. With further fall in P(Ao2), [unk]V(E) increases steeply and the slope of the curve approaches infinity at a tension of 30-40 mm Hg. For purposes of quantitation these curves are approximated by a simple hyperbolic function, the parameters of which are evaluated by a least squares fit of the data. The parameter A denotes curve shape such that the higher the value of A. the greater the increase in ventilation for a given decrease in P(Ao2) and hence the greater the hypoxic drive. Curves are highly reproducible for each subject and curves from different subjects are similar. In 10 normal subjects at resting P(ACo2), A = 180.2 +/-14.5 (SEM). When P(ACo2) is adjusted to levels 5 mm Hg above and below control in six subjects A = 453.4 +/-103 and 30.2 +/-6.8 respectively. These latter values differed significantly from control (P < 0.05). These changes in curve shape provide a clear graphic description of interaction between hypercapnic and hypoxic ventilatory stimuli. At normal P(ACo2) the [unk]V(E)-P(Ao2) curve has an inflection zone located over the same P(o2) range as the inflection in the oxygen-hemoglobin dissociation curve. This indicated that ventilation might be a linear function of arterial oxygen saturation or content. Studies in four subjects have demonstrated that ventilation is indeed related to arterial oxygen content in a linear fashion. These data suggest, but do not prove, that oxygen tension in chemoreceptor tissue as in part determined by circulatory oxygen delivery may be an important factor in controlling the ventilatory response to hypoxia.