BACKGROUND: Hyperventilation is a frequently used method for inducing hypercarbia in neurosurgical patients. This practice requires careful carbon dioxide monitoring that might be replaced by a less expensive and less invasive alternative to arterial blood gas monitoring. OBJECTIVE: To determine the accuracy of end-tidal carbon dioxide monitoring in hyperventilated neurosurgical patients. METHODS: Nineteen adult patients requiring hyperventilation for the reduction of intracranial pressure following head injury or neurosurgery were enrolled from the surgical intensive care unit of a level I trauma center. A correlation design was used to compare arterial carbon dioxide tensions and end-tidal carbon dioxide measurements during specific periods; secondary analysis with bias and precision estimates was performed. Also, changes in arterial carbon dioxide tensions were compared with simultaneous changes in end-tidal carbon dioxide values. RESULTS: End-tidal carbon dioxide values showed a moderately acceptable correlation with arterial blood gas measurements. However, changes in end-tidal carbon dioxide values failed to correlate with simultaneous changes in arterial carbon dioxide tension measures. Bias and precision measures confirmed these findings. CONCLUSION: In this patient sample, changes in end-tidal carbon dioxide values did not accurately reflect changes in arterial carbon dioxide tension levels in the intensive care setting. Further technological advances in noninvasive carbon dioxide monitoring may lead to a significant cost savings over traditional arterial blood gas analysis.