Although nitrous oxide (N2O) should theoretically increase the severity of venous air embolism (VAE), data confirming this hazard in clinical situations are not available. The effect of 50% N2O on the incidence and severity of VAE and on the emergence time from anesthesia was evaluated in 300 neurosurgical patients operated upon while in the sitting position. Of these, 110 patients underwent craniectomy for posterior fossa pathology and 190 patients underwent cervical spine surgery (CSS). Patients were randomized to receive either 50% N2O in oxygen (O2) (N2O group) or O2 (no-N2O group) as part of an isoflurane-fentanyl-based anesthetic. In patients in the N2O group, N2O administration was discontinued immediately upon Doppler-detection of VAE and was reinstituted in not less than 30 min after resolution of the episode. The incidence of Doppler-detected VAE was significantly greater in the craniectomy group than the CSS group (43% vs. 7%, respectively; P less than 0.001). N2O had no effect on the incidence of VAE or the severity of VAE as judged by the magnitude of the reduction in blood pressure during hemodynamically significant episodes of VAE, the volume of gas aspirated from the right atrial catheter during episodes of VAE, or the magnitude of the decrease in end-tidal carbon dioxide tension during episodes of VAE. Hemodynamically significant episodes of VAE (i.e., episodes associated with a reduction in systolic blood pressure of greater than or equal to 15 mmHg) occurred in 17 of the 61 patients experiencing VAE (28%) and was not different between the N2O and no-N2O groups. Similarly, hemodynamically significant episodes of VAE (n = 18) accounted for 15% of all episodes of VAE (n = 118) and was not different between the N2O and no-N2O groups. Emergence time was not significantly different between the N2O and no-N2O groups, with mean times of 2 +/- 6 and 3 +/- 7 min (+/- SD), respectively. Emergence time was significantly longer in the craniectomy group than in the CSS group (5 vs. 1 min, respectively; P less than 0.001). Within the craniectomy group, the incidence of Doppler-detected VAE was significantly less in patients with previous surgery at the operative site (21%) compared to patients without previous surgery at the operative site (47%). Postoperatively, no complications could be related to the use of N2O or directly attributed to the occurrence of VAE.(ABSTRACT TRUNCATED AT 400 WORDS)