Sedation is frequently undertaken for CT in children. Various ward-administered regimens are often successful with modern, fast CT machines, as the procedure itself usually takes only a few minutes, even when further images are required because of excessive movement. Furthermore, the patient is comparatively accessible in the CT machine, and standard monitoring, which is often present in the CT room for use with ventilated emergency patients, can be readily applied to the child. MRI takes much longer than CT, and any movement usually degrades all images of a particular sequence. Orally-administered sedation regimens are therefore more likely to fail. Most 1.5 T MR scanners in current use give poor access to the patient. Whereas adults may at least have their feet outside the magnet, a small child is entirely enclosed within the bore of the magnet and direct observation of the child often requires an attendant to actually crawl into the magnet. An inadequately sedated child who still occasionally moves will result in repeated acquisitions and thus waste time. Running costs of MRI equipment are in excess of £500 per hour. There may be a temptation to give more sedation, either by the staff in the MRI unit or a request to the referring clinicians, which potentially wastes more time. The temptation to give ``a little intravenous valium'' on top of the pre-existing sedation is particularly dangerous unless the administrator has complete con®dence in his ability to perform full resuscitation, and has all the equipment readily to hand. Oral regimens have other problems. Not only may they be unsatisfactory because of patient movement, but they may result in a patient who is deeply unconscious, especially if supplements have been given in an attempt to improve the sedation. A deeply sedated patient results in a conscious level that would result in gastric lavage, administration of antidotes and hospital admission if the child had presented to the A&E department. Such a conscious level is not only dangerous because of aspiration of vomit, or hypoxia from respiratory depression or airway obstruction, but also because the head and body movement associated with exaggerated respiratory effort of a partially obstructed airway can result in even further image degradation. Patient safety has been considered to be of paramount importance by anaesthetists for many years. The introduction of rigorous standards of training, practice and, most important of all, patient monitoring during anaesthesia has resulted in a signi®cant reduction in anaesthetic risk within the United States [1]. In the UK, anaesthetists have moved from the highest risk band and are now in the same risk band (Band 2) as radiologists for indemnity premiums with the Medical Defence Union. It is of paramount importance that any risk associated with a procedure, especially if a diagnostic rather than a life-saving intervention, is minimized. In addition, recent much publicized events in which patients have come to harm suggest that an unexpected death during diagnostic MRI would result in loss of public con®dence. There are enough anecdotes of incidents occurring in MRI units around the country, few of which enter the medical literature, to suggest that the MRI unit is a clinical area with considerable potential for medical accidents. Anaesthetists are in an ideal position to safely administer sedation. When intravenous sedation is administered by someone other than an anaesthetist, that person should retain two-way communication with the patient. However, this light level of sedation does not allow satisfactory MRI in a child. A deeper level of sedation, producing tranquillity and lack of movement, is needed despite the arousal produced by the noise of the MR machine. In the specialist paediatric-only hospitals, a large proportion of the patients passing through the MRI unit will need some form of intervention to facilitate the procedure. It is universally considered impractical to have an anaesthetist available for every MR session, so the majority of children are given either a ward-administered regimen or sometimes sedation supervised by the radiologist. A year-long audit of CT and MRI of children at a large centre in the US, involving a total of 922 children, of whom 530 underwent MRI with sedation, showed a small but worrying incidence Received 22 March 2000 and accepted 22 March 2000. The British Journal of Radiology, 73 (2000), 575±577 E 2000 The British Institute of Radiology