Clinical use of desflurane in neuroanesthesia remains under debate. Comparison of dose-dependent vasodilatory properties between desflurane and isoflurane, the more traditional volatile agent for clinical neuroanesthesia, requires equianesthetic dosing of the agents. Reproducible neurophysiological measurements of the level of anesthesia in an individual, e.g. the A-line autoregressive index (AAI), can be used for an equipotent dosage of two volatile agents in the same individual. Desflurane and isoflurane, in randomized order, were titrated to a stable AAI level of 15-20 in 18 ASA I or II patients. The mean flow velocity (Vmca) and pulsatility index (PI) in the middle cerebral artery were then measured with transcranial Doppler at an end-tidal CO(2) concentration of 4.4%. For desflurane Vmca was 11% higher [95% confidence interval (CI), 5-18%; P = 0.0020] and PI was 13% lower (95% CI, 3-23%; P = 0.0083) than for isoflurane. The mean arterial blood pressure did not differ between the agents. The fraction of MAC necessary for the intended AAI level was 35% lower (95% CI, 20-49%; P = 0.00016) with desflurane than with isoflurane. Desflurane was associated with more cerebral vasodilation than isoflurane at the same depth of anesthesia, as indicated by the AAI. This attributes further reason for caution in the use of desflurane in clinical neuroanesthesia. The difference between desflurane and isoflurane in the MAC fractions required for the same AAI level confirms the limitations of MAC in defining the level of anesthesia.