We tested the hypothesis that left tracheal pulse oximetry (SpO(2)) is more accurate than finger SpO(2) when compared with oxygen saturation from arterial blood samples (SaO(2)) in anesthetized patients with normal thoracic anatomy. We also tested the hypothesis that tracheal oximetry readings are primarily derived from the tracheal mucosa. We studied 20 hemodynamically stable, well oxygenated, anesthetized patients with normal anatomy (ASA physical status I-III, 18-80 yr old). A single-use pediatric pulse oximeter was attached to the left lateral surface of a tracheal tube cuff. Tracheal and finger SpO(2) (dominant index finger), and SaO(2) (nondominant radial artery) were taken with the intracuff pressure at 0-60 cm H(2)O. Tracheal SpO(2) was the same as SaO(2) at an intracuff pressure of 10-60 cm H(2)O, but was less when the intracuff pressure was zero (P<0.0001). Tracheal SpO(2) was higher than finger SpO(2) at an intracuff pressure of 10-60 cm H(2)O (all: P <0.001), but was lower when the intracuff pressure was zero (P< 0.0001). SaO(2) was always higher than finger SaO(2) (P<0.0001). Tracheal SpO(2) was lower at an intracuff pressure of zero (P< 0.0001), but was otherwise similar over the range of intracuff pressures. SaO(2) and finger SpO(2) did not vary with intracuff pressure. Tracheal SpO(2) agrees more closely with SaO(2) than finger SpO(2) at an intracuff pressure of 10-60 cm H(2)O (mean difference < 0.2%). We conclude that left tracheal SpO(2) is feasible and provides similar readings to arterial blood samples and more accurate readings than finger oximetry in hemodynamically stable, well oxygenated, anesthetized patients with normal thoracic anatomy. Tracheal oximetry readings are not primarily derived from the tracheal mucosa. The technique merits further evaluation.