The effect of oral premedication was studied in a double-blind, randomised trial of 200 children undergoing day-stay anaesthesia. Midazolam 0.25 mg/kg, midazolam 0.5 mg/kg, diazepam 0.5 mg/kg or a placebo was given orally one hour prior to anaesthesia. Patient state was assessed at nine stages, from administration of the premedication up to and including induction of anaesthesia, using a four-point behavioural scale. Patient state was also assessed postoperatively in the recovery area and the day-stay ward. There was no difference between the four groups until induction of anaesthesia. At this stage 82% of children were either asleep or awake and calm. Patients who received midazolam 0.5 mg/kg were more likely to be asleep or awake and calm at induction rather than other groups (P = 0.05). Children receiving midazolam 0.5 mg/kg or diazepam 0.5 mg/kg slept longest in the post anaesthetic recovery room (P less than 0.005), and spent most time there (P less than .005). There was no difference between groups in the length of time spent in the day-stay ward or in the number of overnight admissions. The study shows that a high proportion of unsedated children are calm at induction of anaesthesia and that oral midazolam is an effective premedication in children for day-stay anaesthesia.