Pediatric obstructive sleep apnea (OSAS) is characterized by partial airway obstruction, alveolar hypoventilation, and elevated arterial CO2 (Pa-CO2). Thus, a reliable, practical method of estimating CO2 is needed for pediatric polysomnography. Therefore, we measured both transcutaneous CO2 (Ptc(CO2)) and end-tidal CO2 (PET(CO2)) in 15 pediatric polysomnographic evaluations. Sleep state, the highest Ptc(CO2) and the highest PET(CO2) were recorded for 5,159 thirty-second epochs. Although Ptc(CO2) and PET(CO2) were available for 78.5 and 73.0% of epochs, respectively, at least one estimator was available for 92% of the epochs. One infant who would not tolerate a nasal sampling catheter had no PET(CO2) data. For 13 of 14 studies there was a relatively constant difference between Ptc(CO2) and PET(CO2). The difference between Ptc(CO2) and PET(CO2) was within 4 mm Hg in 63.9% of 3,072 epochs. Across 14 studies, mean Ptc(CO2) exceeded mean PET(CO2) by 2.8 +/- 3.0 mm Hg, and it was within 4 mm Hg in 10 studies. In three subjects, PET(CO2) was intermittently or consistently less than Ptc(CO2) because of tachypnea, increased physiologic dead space, or severe partial airway obstruction; in one subject Ptc(CO2) exceeded PET(CO2) for undetermined reasons during one electrode application. The results of this study indicate that Ptc(CO2), as well as PET(CO2), should be measured during pediatric polysomnography. By utilizing both Ptc(CO2) and PET(CO2) there was a 70% reduction in the number of epochs that could not be assessed for hypoventilation. For an individual subject or electrode application there was a constant, and usually close, relationship, between Ptc(CO2) and PET(CO2). Ptc(CO2) monitoring was particularly useful for children who would not tolerate a nasal sampling tube and for those with moderate to severe partial airway obstruction, tachypnea, or increased physiologic dead space in whom PET(CO2) underestimated Ptc(CO2).