Arterial PCO2 (PaCO2) can be continuously and noninvasively estimated by monitoring peak expired CO2 tension (PpeCO2). The practice of calibrating the estimate by an initial measurement of PaCO2 assumes that the difference in PCO2 tension between arterial blood and expired gas P(a-pe)co2 remains constant. We examined the stability of P(a-pe)CO2 during anesthesia in 15 patients undergoing major surgery. Mean P(a-pe)CO2 values ranged from 0.8–7.9 torr with maximum P(a-pe)CO2 values ranging from 4.5–13.0 ton. Calibration of P(a-pe)CO2 based on a single initial measurement of PaCO2 often over- or underestimated PaCO2. Mean estimated PaCO2 from calibrated P(a-pe)CO2 varied from - 7.9–6.4 torr with extreme estimates of — 12.8–12.3 torr. No consistent correlation was shown between P(a-pe)CO2 and duration of anesthesia, variations in ventilation, blood pressure, blood-gas tensions, PpeCO2 or temperature. We conclude that estimation of PaCO2 by monitoring PpeCO2 is not invariably reliable.