A study was undertaken to determine the pattern of end-tidal carbon dioxide (ETCO2) changes during asphyxia-induced cardiac arrest in a pediatric canine model. Eleven intubated, anesthetized, paralyzed dogs (mean age, 4.1 mo; mean weight, 5.5 kg) were used. Asphyxia was induced by clamping the endotracheal tube (ETT) and discontinuing ventilation. Cardiac arrest ensued a few minutes later, after which closed-chest cardiopulmonary resuscitation (CPR) and ventilation were initiated. The ETCO2 level was recorded at baseline and every minute during CPR. Mean baseline ETCO2 was 31.9 mm Hg. The initial ETCO2 immediately after unclamping the ETT (mean, 35 mm Hg) was higher than subsequent values (mean, 12.4 mm Hg; P < .001). There was a sudden increase in ETCO2 to a mean of 27.0 mm Hg at or just before return of spontaneous circulation (ROSC) in all 11 cases (P < .01). During CPR, ETCO2 levels were initially high, decreased to low levels, and increased again at ROSC. This pattern, not previously described, is different from that observed in animal and adult cardiac arrest caused by ventricular fibrillation, during which ETCO2 decreases to almost zero after the onset of arrest, begins to increase after the onset of effective CPR, and increases to normal levels at ROSC. In this model of asphyxial arrest, continued cardiac output prior to arrest allows continued delivery Of CO2 to the lungs, resulting in higher alveolar CO2; this, in turn, is reflected as increased ETCO2 once ventilation is resumed during CPR. Further study is needed to determine whether the pattern Of ETCO2 changes can be used prospectively to define the etiology of cardiac arrest.