The authors performed a randomised, prospective trial in which one junior anaesthetist administered gaseous induction of anaesthesia to 50 unpremedicated children with either isoflurane or halothane in nitrous oxide and oxygen. Arterial oxygen saturation and the electrocardiogram were monitored and the incidence of complications noted. Desaturation below 85% occurred in six children, but only with isoflurane. The incidences of complications and desaturation events did not alter throughout the 25 isoflurane inductions. Coughing, movement, laryngospasm and sinus tachycardia occurred more frequently with isoflurane. Isoflurane inductions took longer (7.9 as compared with 5.4 minutes, p less than 0.001) and had 4.25 times the number of complications.