PETCO2 is often used as an estimate of PaCO2, with the understanding that PaCO2 usually exceeds PETCO2. During intraoperative craniotomies, because hyperventilation is used to therapeutically lower intracranial pressure, the difference between PaCO2 and PETCO2 (P(a-ET)CO2) has therapeutic implications. The P(a-ET)CO2 was hypothesized to be stable during craniotomies with relatively short-term monitoring and controlled cardiorespiratory variables. Thirty-five patients undergoing elective craniotomies were studied. Arterial blood gases (with PaCO2) were measured after induction of general anesthesia, after cranium opening prior to dural incision, and at start of closure; PETCO2 was simultaneously determined with infrared capnometry. The PaCO2 was 31.9 +/- 3.9 mm Hg (range, 24.8-46.7) (values are mean +/- SD) and PETCO2, 24.7 +/- 3.8 mm Hg (range, 16-34), with a P(a-ET)CO2 of 7.2 +/- 3.3 mm Hg (of 126 comparisons, range was -1.2-17.3). There was no correlation of P(a-ET)CO2 with blood pressure, heart rate, respiratory volumes, airway pressures, or inspired oxygen concentration. There was a significant positive correlation between PaCO2 and PETCO2 (r = 0.632, slope = 0.609) and P(a-ET)CO2 and PaCO2 (r = 0.46, slope = 0.391, P < 0.017, and r2 = 0.22). Although changes in the study population of PaCO2 and PETCO2 correlated statistically (r = 0.818, slope = 0.76, P < 0.001, r2 = 0.669), comparisons in 17 of 35 individuals were not significant. On comparison of subsequent measurements, 18.4% of changes in PaCO2 and PETCO2 (although sometimes small) were in opposite directions. P(a-ET)CO2 did not change with time. The PETCO2 does not provide a stable reflection of PaCO2 in many patients undergoing craniotomies.