Pulse oximetry is used extensively during exercise stress-testing in the clinical and sports medicine settings. There are few validation studies to assess the appropriateness of using pulse oximetry under conditions of potentially compromised peripheral blood flow. To study the accuracy of pulse oximetry during severe exercise stress, 10 athletes undertook 3 bouts of exhaustive exercise; once at an intensity requiring V̇O2max (max), once at 115% of V̇O2max (Smax), and once at Smax while FIO2 was increased to 0.30. The results indicate relatively large underestimations occur when pulse oximetry is used to estimate %SaO2 during exercise, when compared to the criterion samples of gas analysis in arterial blood. These differences were exacerbated as the exercise intensity increased from a mean (± SE) difference of 2.9 ± 0.7 %SaO2 at max to 4.6 ± 0.7 %SaO2 at Smax. Breathing a higher FIO2 reversed the hypoxemia that occurred during the normoxic exercise, however, pulse oximetry measurements failed to detect this alteration in %SaO2. Estimates of oxygen saturation during severe exercise using pulse oximetry should be viewed with caution, as potentially large errors may occur.