Objective. To assess the correlation and accuracy of end-tidal PCO2 (PetCO2) sampled via nasal cannulae in pediatric patients by comparison to the criterion standard PaCO2, and to identify sources of error during PetCO2 monitoring via nasal cannulae.Methods. PetCO2 was monitored continuously by sampling end-tidal gas through nasal cannulae that had been designed and manufactured for this purpose in spontaneously breathing children undergoing conscious or deep sedation during either cardiac catheterization (n = 43) or critical care (n = 54). When both the capnographic wave form and the PetCO2 value had been stable for at least 10 minutes, the PetCO2 value was recorded while blood was drawn from an indwelling arterial line for PaCO2 measurement. The effects of age, weight, respiratory rate, oxygen delivery system, airway obstruction, mouth breathing, and cyanotic heart disease were evaluated by linear regression analysis and calculation of absolute bias (PaCO2-PetCO2).Results. Mouth breathing, airway obstruction, oxygen delivery through the ipsilateral nasal cannula, and cyanotic heart disease adversely affected accuracy. In patients without those factors, PetCO2 correlated well with PaCO2 (R2 = 0.994), and absolute bias was 3.0 ± 1.8 mmHg.Conclusions. Several factors — some controllable and all recognizable — affect the accuracy of PetCO2 monitored via nasal cannulae in pediatric patients. When these factors are not present, PetCO2 correlates well with PaCO2 and appears to be a useful monitor of ventilatory status during conscious or deep sedation.