This study was designed to evaluate the clinical characteristics of three induction techniques using sevoflurane in children scheduled for tonsillectomy: incremental induction with sevoflurane(2,4,6,7%) in 100% O2 (group IC-O2; n=23); induction with high concentration of sevoflurane in 100% O2 (group HC-O2; n=22); and induction with high concentration of sevoflurane in a mixture of O2:N2O(50:50) (group HC-N2O; n=20). Induction was well accepted and well tolerated in most children. The addition of nitrous oxide resulted in faster loss of consciousness (P< 0.001) compared to the other induction techniques and in a tendency for reduced excitement compared with the same rapid technique without nitrous oxide (P=0.053). Time to tracheal intubation was identical in the three groups and intubation conditions were scored as good in most children. During the early induction phase, an increase in SAP and HR was observed in the three groups. Changes were maximal at two min after the beginning of induction in the three groups. SAP and HR values were back to baseline values at the time of tracheal intubation. In conclusion, the addition of nitrous oxide to a high sevoflurane concentration decreases the time to loss of eyelash reflex, tends to reduce the incidence of excitement and is not associated with an increased incidence of respiratory complications even in patients with obstructive airway.