The ability of continuous infusions of opioids to control hypertension at the end of neurosurgical procedures without compromising prompt emergence was studied in patients undergoing craniotomy for supratentorial tumours. Four infusion regimens were compared in a randomized double-blind fashion; three of alfentanil and one of fentanyl. Low-dose alfentanil was administered to nine patients (35.1 micrograms.kg-1 then a continuous infusion of 16.2 micrograms.kg-1.hr-1); mid-dose alfentanil to eight patients (70.2 micrograms.kg-1 then 32.4 micrograms.kg-1.hr-1); high-dose alfentanil to eight patients (105.3 micrograms.kg-1 then 48.6 micrograms.kg-1.hr-1). Eight additional patients were given fentanyl (8.3 micrograms.kg-1 then 1.6 micrograms.kg-1.hr-1). Using published values for the pharmacokinetic variables of alfentanil and fentanyl, modelling predicted stable concentrations of 60, 120, 180 ng.ml-1 for the alfentanil infusion regimens respectively and 2 ng.ml-1 with the fentanyl regimen. Maintenance anaesthesia comprised the opioid infusion, 50% N2O in O2 and isoflurane titrated to control mean arterial pressure (MAP) within 20% of ward MAP. Isoflurane was discontinued after closure of the dura. Nitrous oxide was discontinued at the same time as reversal of neuromuscular blockade. The opioid infusion was discontinued with closure of the galea. A greater time-averaged isoflurane concentration was required to control MAP within the prescribed limits in the low alfentanil group (ANOVA; P less than 0.05). The PaCO2 at two, five and 30 min after extubation were not different among groups. The times from discontinuing N2O to eye opening and tracheal extubation were not different. The time to follow commands was longer in the low alfentanil group (P less than 0.05).(ABSTRACT TRUNCATED AT 250 WORDS)