Hyperventilation produces small decreases in blood pressure in normal subjects and larger decreases in patients with autonomic failure. The authors studied the mechanism for this observation by measuring mean arterial pressure (MAP) and arterial blood gas (ABG) changes in eight patients with severe primary autonomic failure after various maneuvers designed to alter PaCO2, PaO2, and pH. Maneuvers included voluntary hyperventilation, breathing a 5% CO2/95% O2 mixture, breathing 12% O2, breathing through a 1 meter tube to increase dead space, breathing 100% O2, and infusion of 120 mEq NaHCO3 over 30 minutes. All maneuvers led to expected changes in ABGs. Voluntary hyperventilation lowered MAP by 23 +/- 4 (p less than 0.01) mmHg but MAP was raised 11 +/- 3 and 7 +/- 1 mmHg by hyperventilation resulting from increasing breathing dead space or from breathing 5% CO2, respectively. Breathing 100% O2 or 12% O2 had no significant effect on MAP, and NaHCO3 infusion raised MAP by 8 +/- 4 (p less than 0.05) mmHg. With all maneuvers, change in MAP correlated with change in PaCO2 (r = 0.72, p less than 0.001) and change in pH (r = -0.57, p less than 0.01) but not with PaO2. Multiple regression analysis showed that only changes in PaCO2 predicted the change in MAP for all maneuvers. The authors conclude that a decrease in PaCO2 causes the observed decreases in MAP with hyperventilation. This most likely represents a direct peripheral vasodilator effect of hypocarbia rather than a reflex or centrally-mediated mechanism since our patient population is characterized by inadequate or absent autonomic cardiovascular reflex responses.