We studied prospectively the effects of the technique of tracheal extubation on arterial haemoglobin oxygen saturation (SpO2) in 120 ASA I-III children, mean age 5.3 (range 0.25-16.9) yr. At completion of surgery, tracheal extubation was performed when spontaneous ventilation had resumed, children were fully awake and SpO2 was 99-100%. Children were allocated randomly to receive a single lung inflation manoeuvre with 100% oxygen before tracheal extubation (group I; n = 59) or to have the tracheal tube removed while applying suction through the tube (group S; n = 61). SpO2 was monitored during the first 5 min after tracheal extubation in the operating room. Supplementary oxygen was given if SpO2 decreased to less than 92%. The time between tracheal extubation and decrease in SpO2 to 92% (T92) was recorded. Children in group S required oxygen administration more frequently after tracheal extubation than those in group I (65.6% vs 45.8%; P = 0.04), and had a three-fold shortening of T92 (mean 25 (SD 19) s vs 85 (63) s; P = 0.0001). These effects were more pronounced in children less than 4 yr of age compared with older children. We conclude that tracheal extubation greatly impaired oxygenation and therefore administration of oxygen was appropriate. This impairment was more marked when suction was used, and in young children. Lung inflation with 100% oxygen before removal of the tracheal tube is advised before routine tracheal extubation in children.