BACKGROUND: A reduction in the arterial partial pressure of CO2 (PaCO2) leads to a rapid reduction in cerebral blood flow (CBF). However, despite continuing hypocapnia there is secondary recovery of CBF over time as a result of increases in lactic acid production. Hyperoxia is thought to modulate the production of lactic acid. This study examined the kinetics of middle cerebral artery flow velocity (MCA FV) reduction during hyperventilation, and its modulation by hyperoxia. METHODS: Cerebral blood flow was assessed using transcranial Doppler ultrasound in nine healthy, awake human volunteers. Subjects were ventilated, via a mouthpiece, to achieve a stable end-tidal CO2 (PETCO2). After a 20-min baseline period the minute volume on the ventilator was passively increased by approximately 20% to reduce PETCO2 by 0.75-1 kPa. After a 10-min stabilization period the new PETCO2 level was maintained at a constant level for 20 min, and MCA FV recovery was measured during this 20-min period. Subjects undertook the protocol breathing air and breathing 100% oxygen. RESULTS: The PETCO2 level was (mean +/- SD) 4.9 +/- 0.4 kPa (normoxia baseline), 4.0 +/- 0.3 kPa (normoxia hyperventilation), 4.6 +/- 0.4 kPa (hyperoxia baseline) and 3.9 +/- 0.4 kPa (hyperoxia hyperventilation). CO2 reactivity was significantly lower with normoxia than hyperoxia (16.5 +/- 3.8 vs. 21.2 +/- 4.6 % kPa-1; P< 0.05). Middle cerebral artery FV recovery was significantly more rapid with normoxia than hyperoxia (0.23 +/- 0.17 vs. 0.08 +/- 0.1 % baseline min-1; P< 0.01). CONCLUSIONS: Our results suggest that cerebral hemodynamic responses to moderate hyperventilation are different in normoxic and hyperoxic conditions. Clinical assessment of CO2 reactivity and CBF recovery during hyperventilation should take the degree of arterial oxygenation into account.