BACKGROUND: During craniotomy operations, the PaCO2 has therapeutic implications because hyperventilation is often used to lower intracranial pressure. PETCO2 is often used as an estimate of PaCO2, with the assumption that P(a-ET)CO2 is relatively constant. To clarify the relationship between PaCO2 and PETCO2, sixty patients undergoing elective craniotomies were studied. METHODS: Arterial blood gases were measured from 30 minutes after endotracheal intubation to skin closure at an interval of 30 minutes in thirty patients, and at random interval in another thirty patients. PETCO2 was simultaneously determined with infrared capnography(Datex AS/3TM, Filand). RESULTS: The PaCO2 was 31.7+/-3.0 mmHg and PETCO2, 26.3+/-2.5 mmHg, with a P(a-ET)CO2 of 5.5+/-2.7 mmHg(n = 431, range between 0-13.5). There was a significant positive correlation between PaCO2 and PETCO2(r = 0.537, slope = 0.440, P<0.001) and between P(a-ET)CO2 and PaCO2(r = 0.625, slope = 0.555, P<0.001). Although changes in the pooled data of PaCO2 and PETCO2 correlated statistically, comparisons in 43 of 60(71.6%) individuals were not correlated. On comparisons of subsequent measurements, 17.0% of changes in PaCO2 and PETCO2 were in opposite directions. P(a-ET)CO2 had a tendency to increase with time during surgery(slope = 0.0082), but there was no statistically significant difference between the measurements. CONCLUSION: The PETCO2 measured with infrared capnography does not provide a stable reflection of PaCO2 in many patients undergoing craniotomy. Therefore, we concluded that capnography must be used in conjuction with arterial blood gas measurements for monitoring the respiratory acid-base status of mechanically ventilated neurosurgical patients undergoing craniotomy.