Accidental bronchial intubation was examined in the first 3947 cases reported to the Australian Incident Monitoring Study and was found to have accounted for 154 (3.7%) of the total incidents reported. Most incidents were detected in the operating theatre (93.5%) and during maintenance of anaesthesia (77.9%), by unexplained oxygen desaturation alone (63.6%). Capnography remained normal or unremarkable during 88.5% of the episodes. One-third of cases were associated with head or neck surgery and possible flexion of the patient's head. A RAE tube was used in 20% of incidents, a greater frequency than occurred in the study overall. A third party was implicated in 36 (23.4%) of cases. Ninety per cent of cases were considered preventable. Major morbidity occurred in three cases and unplanned intensive care admission was required in a further five. Almost two-thirds (61.1%) of the incidents might have been avoided by the proposed markings on the tracheal tube. We conclude that when arterial desaturation occurs at any stage during anaesthesia the possibility of bronchial intubation must be considered. Asymmetrical ventilation may be difficult to detect clinically and in most cases there is no change in capnography.