The CO2 measured by a heated skin surface electrode is greater than arterial CO2. In this study we (1) determined the magnitude of this difference retrospectively, (2) adjusted the calibration of the heated transcutaneous CO2 electrode to reflect this difference, and (3) tested the validity of the calibration procedure prospectively. The retrospective study consisted of 252 simultaneous arterial and transcutaneous measurements on 38 infants and children (age range, 1 day to 6 yr). A linear regression analysis showed PtcCO2 = 1.61 X PaCO2 - 0.01 mmHg, r = 0.90 (p less than 0.0001). Because the intercept was virtually zero, we subsequently changed the calibration procedure by dividing the transcutaneous electrode reading by 1.61. The prospective study consisted of 269 more simultaneous arterial and transcutaneous measurements from 57 consecutive infants and children (age range, 1 day to 18 yr) using this calibration procedure. Regression analysis showed PtcCO2 = 1.02 X PaCO2 - 2.3 mmHg, r = 0.96 (p less than 0.0001). Thus, we found that the transcutaneous CO2 electrode, using this technique, reliably predicted the arterial CO2.