Five dogs were anesthetized, intubated, and ventilated with various mixtures of oxygen, nitrogen, and carbon monoxide. Each dog was monitored with arterial and pulmonary artery catheters, a transcutaneous PO2 analyzer, and two pulse oximeters. An IL-282 Co-oximeter was used to periodically measure arterial oxyhemoglobin (O2Hb) and carboxyhemoglobin (COHb) as percentages of the total hemoglobin. The PaO2, PaCO2, and pHa were measured in the same blood specimens using standard electrodes. When the inspired oxygen concentration was reduced in the absence of COHb, the pulse oximeter saturation (SpO2) estimated O2Hb with reasonable accuracy. COHb levels were then varied slowly from 0-75% in each dog. As the COHb level increased and oxyhemoglobin decreased, both pulse oximeters continued to read an oxygen saturation of greater than 90%, while the actual O2Hb fell below 30%. In the presence of COHb, the SpO2 is approximately the sum of COHb and O2Hb, and, thus, may seriously overestimate O2Hb. The pulse oximeter, as the sole indicator of blood oxygenation, should, therefore, be used with caution in patients with recent carbon monoxide exposure. On the other hand, transcutaneous PO2 falls linearly as COHb increases, and reaches about one-fifth of its initial value at the highest COHb levels despite the maintenance of constant arterial PO2.