BACKGROUND The fraction of inspired oxygen (F(IO(2))) is quoted for different oxygen delivery systems, but variations in inspiratory flow and tidal volume make precise measurement difficult. We developed a reliable method of measuring the effective F(IO(2)) in patients receiving supplemental oxygen. METHODS Ten subjects with chronic hypoxemia breathed through a mouthpiece with a sampling probe connected to a mass spectrometer. Four of the 10 subjects had transtracheal catheters that allowed direct sampling of tracheal gas. We used oxygen concentrations of 47% and 97%, and flow rates between 1 L/min and 8 L/min. We also compared oxygen delivery via nasal cannula and transtracheal catheter. Effective F(IO(2)) was derived from plots of the fractional concentrations of carbon dioxide versus oxygen. RESULTS We found excellent correlation between the effective F(IO(2)) values from tracheal and oral sampling (r = 0.960, P < .001). With 97% oxygen via nasal cannula, effective F(IO(2)) increased by 2.5% per liter of increased flow (P < .001); effective F(IO(2)) reached 32.7% at 5 L/min while P(aO(2)) increased by 12 mm Hg per liter of increased flow. In 4 subjects with a transtracheal catheter, effective F(IO(2)) increased 5.0% (P < .001) per liter of increased flow, and P(aO(2)) increased by 13 mm Hg per liter of increased flow, whereas in the same 4 subjects using nasal cannula for oxygen delivery, P(aO(2)) increased by only 6 mm Hg per liter of increased flow. CONCLUSIONS Exhaled gas sampled at the mouth accurately reflected the effective F(IO(2)) in the trachea. In relation to inspired oxygen flow, the effective F(IO(2)) was lower than is conventionally thought. Compared to nasal cannula, transtracheal catheter approximately doubled the effective F(IO(2)) at a given flow rate. Accurate knowledge of F(IO(2)) should aid clinicians in managing patients with acute and chronic lung diseases.