We tested the performance of transcutaneous oxygen monitoring (TcPO2) and pulse oximetry (tcSaO2) in detecting hypoxia in critically ill neonatal and pediatric patients. In 54 patients (178 data sets) with a mean age of 2.4 years (range 1 to 19 years), arterial saturation (SaO2) ranged from 9.5 to 100%, and arterial oxygen tension (PaO2) from 16.4 to 128 mmHg. Linear correlation analysis of pulse oximetry vs measured SaO2 revealed an r value of 0.95 (p less than 0.001) with an equation of y = 21.1 + 0.749x, while PaO2 vs tcPO2 showed a correlation coefficient of r = 0.95 (p less than 0.001) with an equation of y = -1.04 + 0.876x. The mean difference between measured SaO2 and tcSaO2 was -2.74 +/- 7.69% (range +14 to - 29%) and the mean difference between PaO2 and tcPO2 was +7.43 +/- 8.57 mmHg (range -14 to +49 mmHg). Pulse oximetry was reliable at values above 65%, but was inaccurate and overestimated the arterial SaO2 at lower values. TcPO2 tended to underestimate the arterial value with increasing PaO2. Pulse oximetry had the best sensitivity to specificity ratio for hypoxia between 65 and 90% SaO2; for tcPO2 the best results were obtained between 35 and 55 mmHg PaO2.