To determine how closely peak expired PCO2 measured by mass spectrometry reflects arterial PCO2 during anesthesia and what variables contribute to a difference between the two measurements, peak expired CO2, arterial PCO2, and other physiologic variables were measured simultaneously. There was a significant correlation between peak expired and arterial PCO2 (p less than 0.001). The difference between temperature-corrected arterial and peak expired PCO2 was related to the presence or absence of lung disease, age, A.S.A. class, and systolic blood pressure. No significant relationship was seen between the temperature-corrected difference and duration of anesthesia, diastolic pressure, expired O2 concentration, or anesthetic agent. Mean temperature-corrected arterial PCO2 exceeded mean peak expired PCO2 by 1.7 torr in all patients, 0.8 torr in patients without lung disease, and 3.3 torr in patients with lung disease. In patients in whom more than one arterial sample was obtained, initial differences correlated significantly with subsequent differences (p less than 0.001).