We sought to compare effects of remifentanil- and fentanyl-based anesthesia on the morphology of somatosensory evoked potentials (SSEPs) and speed of recovery from anesthesia. Forty-one patients undergoing spinal surgery and requiring intraoperative monitoring of SSEPs were randomized into two groups. In Group 1, anesthesia was induced with sodium thiopental and maintained with fentanyl, 50% nitrous oxide in oxygen, and 0.5%--0.75% isoflurane. In Group 2, anesthesia was induced with sodium thiopental and maintained with remifentanil, 50% oxygen in air, and 0.5%--0.75% isoflurane. The variables compared included hemodynamic changes during the induction and intubation, the interval from the end of anesthesia to extubation, intraoperative blood loss and fluid administration, and changes in latency and amplitude of the P37--N45 component of posterior tibial nerve somatosensory evoked potentials and the N20--P24 component of median nerve somatosensory evoked potentials. The two groups were matched for demographics, ASA physical status, and duration of surgery. Hemodynamic profiles after the induction and intubation were similar. There were significant differences between groups in time intervals from the end of anesthesia to extubation (15.3 +/- 12.8 vs 5.3 +/- 2.3 min; P = 0.0001) and ability to follow verbal commands (14.6 +/- 11.9 vs 4.5 +/- 2.4 min; P = 0.0001), with the Remifentanil group showing earlier recovery. Variability (coefficient of variation) of P37--N45 latency was greater (0.026 vs 0.014; P = 0.001) in the Fentanyl group.