Forty-two children (aged 27–16 years) were randomly assigned to receive either sevoflurane (n=21) or halothane (n=21) in nitrous oxide/oxygen. After premedication with midazolam, anaesthesia was induced by facemask and the anaesthetic concentration was increased until loss of eyelash reflex (sevoflurane, 6%; halothane, 2.5%). Thereafter, 1–1.5 MAC of the inhalational agents were maintained until skin closure. Intra-operative analgesia was provided either by intermittent intravenous (i.v.) bolus doses of fentanyl (2–3 μg kg−1) or by a regional blockade. Induction was smooth and the time to loss of eyelash reflex was slightly shorter with sevoflurane than with halothane, the difference not quite reaching statistical significance (P=0.06). In both groups, heart rate remained stable and blood pressure decreased significantly during induction. Haemodynamic parameters remained stable during anaesthetic maintenance; no cardiac dysrhythmias were observed. Emergence time with sevoflurane was 12.9 min vs. 16.3 min with halothane, but this difference was not statistically significant. It is concluded that sevoflurane is as suitable for paediatric patients as halothane. The slightly faster emergence time offered by sevoflurane over halothane was of no clinical significance in the present study.