Arterial blood gas measurements (ABGs) are the most common tests ordered in an ICU. ABG utilization in a surgical ICU over a 1-year period (September 1, 1987-October 31, 1988) was evaluated to identify factors that might help reduce overutilization. A total of 842 admissions comprising 2,381 patient days were reviewed. ABGs were the most commonly ordered test (mean of 4.8/patient/day). Patients with arterial lines (A-lines) had more ABGs drawn than those who did not regardless of the value of PaO2 (p less than 0.01), PaCO2 (p less than 0.01 except for PaCO2 greater than 55), APACHE II score (p less than 0.01), use of ventilators (p less than 0.01), pulse oximeters (p less than 0.01), or a combination of the last two (p less than 0.01). Multivariate analysis demonstrated that the presence of an A-line was the most powerful predictor of the number of ABGs drawn per patient (p less than 0.0001) independent of all other measures of the patient's clinical status such as the use of ventilators, oximeters, and values of PaO2, PaCO2, or the APACHE II score. This suggests that ABGs are being drawn unnecessarily simply because of the presence of an A-line. To reduce the number of ABGs drawn, a policy for specific indications for placement of A-lines and ABG analysis should be adopted.