The normal ventilatory response to the sudden imposition of sustained hypoxia is characterized by an acute increase followed by a modest decline in ventilation. Since subanesthetic concentrations of potent inhalational anesthetics greatly attenuate the acute response, we hypothesized that ventilation might decrease to less than normoxic levels when hypoxia is sustained. We therefore measured the ventilatory response to 20 min of sustained hypoxia (PET(O2) 45 mmHg) at two levels of strict isocapnia-normocapnia (PET(CO2) 1-2 mmHg above resting) and hypercapnia (PET(CO2) 49 mmHg)-in eight healthy male subjects during inhalation of 0.1 MAC isoflurane or carrier gas (control). An abrupt end-tidal step from normoxia to isocapnic hypoxia was induced using a dynamic end-tidal forcing system. Isoflurane and control experiments were performed on separate days; the order of isoflurane and control days and the order of normocapnia and hypercapnia within days were randomized. Subjects were studied while fasted, always at the same time of day, and were required to watch a documentary videotape to minimize differences in level of consciousness. With normocapnia, there was no difference in ventilation at any time between isoflurane and control (prehypoxic 9.6 +/- 1.5 vs. 9.5 +/- 2.6 l/min, peak hypoxic 24.7 +/- 10.4 vs. 26.2 +/- 10.4 l/min, final hypoxic 15.0 +/- 4.4 vs. 15.9 +/- 3.5 l/min; mean +/- SD). With hypercapnia, prehypoxic ventilation increased to the same level for isoflurane and control (24.8 +/- 6.7 vs. 24.8 +/- 9.6 l/min). Although peak hypoxic ventilation was slightly less in isoflurane than in control hypercapnic experiments, this was not significant (49.6 +/- 16.3 vs. 56.5 +/- 24.3 l/min; P = .22). But because ventilation also decreased slightly, but not significantly, more in isoflurane than in control hypercapnic experiments, final hypoxic ventilation was less with isoflurane (40.8 +/- 13.1 vs. 48.5 +/- 16.3 l/min; P < .05). Our data indicate that 0.1 MAC levels of isoflurane do not affect the ventilatory response to sustained normocapnic hypoxia or to normoxic hypercapnia but do reduce the ventilatory response to sustained hypercapnic hypoxia and may therefore attenuate the normal hypoxic-hypercapnic interaction.