We sought to describe cerebrovascular responses to incremental exercise and test the hypothesis that changes in cerebral oxygenation influence maximal performance. Eleven men cycled in three conditions: 1) sea level (SL); 2) acute hypoxia [AH; hypobaric chamber, inspired PO2 (PIO2) 86 Torr]; and 3) chronic hypoxia [CH; 4,300 m, PIO2 86 Torr]. At maximal work rate (Ẇmax), fraction of inspired oxygen (FIO2) was surreptitiously increased to 0.60, while subjects were encouraged to continue pedaling. Changes in cerebral (frontal lobe) (COX) and muscle (vastus lateralis) oxygenation (MOX) (near infrared spectroscopy), middle cerebral artery blood flow velocity (MCA Vmean; transcranial Doppler), and end-tidal PCO2 (PETCO2) were analyzed across %Ẇmax (significance at P < 0.05). At SL, PETCO2, MCA Vmean, and COX fell as work rate rose from 75 to 100% Ẇmax. During AH, PETCO2 and MCA V mean declined from 50 to 100% Ẇmax, while C OX fell from rest. With CH, PETCO2 and COX dropped throughout exercise, while MCA Vmean fell only from 75 to 100% Ẇmax. MOX fell from rest to 75% W? max at SL and AH and throughout exercise in CH. The magnitude of fall in COX, but not MOX, was different between conditions (CH > AH > SL). FIO2 0.60 at Ẇmax did not prolong exercise at SL, yet allowed subjects to continue for 96 ± 61 s in AH and 162 ± 90 s in CH. During FIO2 0.60, COX rose and MOX remained constant as work rate increased. Thus cerebral hypoxia appeared to impose a limit to maximal exercise during hypobaric hypoxia (PIO2 86 Torr), since its reversal was associated with improved performance.