Quality assurance data were collected prospectively for children who were sedated (n = 922) or given general anaesthesia (n = 140) for magnetic resonance imaging (MRI) or computerized tomography (CT). The data included patient characteristics, concurrent medication, adequacy of sedation, adverse events and requirement for escalated care. The quality of scans was evaluated. Reasons for preselection of general anaesthesia included previously failed sedation (28%), potential for failed sedation (32%) and perceived medical risk (14%). Hypoxaemia occurred in 2.9% of sedated children, and was more common in children classified as ASA III or IV. Sedation was inadequate for 16% of children and failed in 7%. Failed sedation was associated with greater age (P = 0.009), higher ASA status (P = 0.04) and use of benzodiazepines as sole sedatives (P < 0.03). More of the children who underwent general anaesthesia were ASA III or IV than sedated children, yet the procedure was successful in all the children who underwent general anaesthesia, with one incident of laryngospasm. Excessive motion was noted in 12% of scans of sedated children and 0.7% of those completed with general anaesthesia. We conclude that sedation of children for MRI and CT is associated with risks of hypoxaemia and of inadequate or failed sedation. These adverse events were more likely to occur in older children, those with a higher ASA status and those in whom benzodiazepines had been used as sole sedatives. For a preselected high-risk group of children, general anaesthesia may make MRI and CT scans more successful with minimal adverse events.