We have studied the effects of sevoflurane on neurological outcome in a rat model of incomplete cerebral ischaemia. After institutional approval, 30 non-fasted male Sprague-Dawley rats (455-555 g) were anaesthetized, the trachea intubated and the lungs ventilated mechanically with isoflurane and 30% oxygen in air. Catheters were inserted into the right femoral artery, both femoral veins and into the right jugular vein for measurement of arterial pressure, drug administration and blood sampling. At completion of surgery, isoflurane was discontinued and the rats were allowed an equilibration period of 30 min according to the following regimens: group 1 (n = 10) received 70% nitrous oxide in oxygen and fentanyl (bolus 10 micrograms kg-1 i.v.; infusion 25 micrograms kg-1 h-1); group 2 (n = 10) received 1.98 vol% sevoflurane in oxygen and air (FIO2 0.3); group 3 (n = 10) received 1.98 vol% sevoflurane in oxygen and air (FIO2 0.3) and 40% glucose (6 ml kg-1 i.p.) 30 min before ischaemia. Ischaemia was produced by combined unilateral common carotid artery ligation and haemorrhagic hypotension to 35 mm Hg for 30 min. Temperature, arterial blood-gas variables and arterial pH were maintained within the physiological range. Plasma glucose concentration was measured before, during and after ischaemia. Neurological deficit was evaluated for 3 days after ischaemia. Neurological outcome was better in sevoflurane anaesthetized animals, regardless of the plasma glucose concentration, compared with nitrous oxide-fentanyl controls. This indicates that differences in plasma glucose concentrations do not account for the cerebral protection seen with sevoflurane.