Emergence delirium, characterised by inconsolable crying, perceptual disturbances and thrashing, occurs in young children during the recovery phase from general anaesthesia. Our aim was to determine whether timing of laryngeal mask airway removal (deeply anaesthetised versus awake) influenced the incidence of emergence delirium in children after tonsillectomy. A single-centre, randomised controlled trial was conducted at Albury Wodonga Health, a regional hospital in Australia. Included patients were two to seven years old, American Society of Anesthesiologists physical status classification 1-2, undergoing elective tonsillectomy (with or without adenoidectomy or grommet insertion) under general anaesthesia. Patients were randomised to have their laryngeal mask removed whilst deeply anaesthetised (in the operating theatre) or after awakening (in the post-anaesthesia care unit (PACU)). Pediatric Anesthesia Emergence Delirium score was determined at 5 and 20 min after eye opening, and frequency of complications (cough, vomiting, excessive salivation, oxygen desaturation and laryngospasm) in the PACU were recorded. Sixty-two patients were randomised to deep laryngeal mask removal and 62 to awake. In the awake versus deep groups, 33 (53%) versus 40 (65%) participants had emergence delirium at 5 min (odds ratio (OR) 0.63, 95% confidence interval (CI) 0.30 to 1.29, P = 0.20). At 20 min, 18 (29%) vs. 19 (31%) participants had emergence delirium (OR 0.93, 95% CI 0.43 to 2.00, P = 0.88). A greater incidence of most PACU complications was observed in the awake versus deep group; cough (24% vs. 8%), vomiting (8% vs. 0%), excessive salivation (23% vs. 8%) and oxygen desaturation (16% vs. 0%). We found no significant difference between the two techniques in terms of preventing emergence delirium. However, other PACU complications were more frequent with awake removal.