In a premedication study involving 135 children, aged 1-10 years, four regimens were investigated: (i) no premedication; (ii) oral trimeprazine tartrate 2 mg/kg, methadone 0.1 mg/kg, droperidol 0.15 mg/kg (TMD); (iii) intramuscular midazolam (Dormicum; Roche) 0.15 mg/kg; and (iv) oral midazolam 0.45 mg/kg. All premedications were given 60 minutes before a standard halothane anaesthetic. No impairment of cardiovascular stability occurred but after premedication the mean oxygen saturation decreased by 1.6% and 1.1%, respectively, in the intramuscular midazolam and TMD groups. Overall, children under 5 years of age behaved less satisfactorily in the holding room and at induction, than those over 5 years (P less than 0.01). Midazolam, intramuscularly and orally, produced more satisfactory behaviour than the other two regimens (P less than 0.05) and, combined with a 70% more rapid recovery than the TMD regimen (P less than 0.05), suggests that oral midazolam is a more effective paediatric premedication agent than placebo or TMD.