The purpose of this study was to evaluate various doses of sublingual midazolam premedication in children. In our prospective, double-blind, placebo-controlled trial, children (n = 102, age range 12 to 129 months) scheduled for day surgery were randomized to receive either midazolam in one of three doses (0.25, 0.5, or 0.75 mg.kg-1) or placebo. Injectable midazolam was mixed with a thick grape syrup and placed under the tongue; the patient was asked to hold it as long as possible before swallowing. Children readily accepted the mixture. Analysing all patients randomized, none of the children receiving placebo vs 28% receiving 0.25 mg.kg-1 (P = 0.02), 52% receiving 0.5 mg.kg-1 (P < 0.001), and 64% receiving 0.75 mg.kg-1 (P < 0.001) of midazolam showed satisfactory sedation (drowsy) at 15 min after administration. Children receiving the two higher doses of midazolam (0.5 and 0.75 mg.kg-1) accepted mask induction willingly, while the group receiving 0.25 mg.kg-1 resembled the placebo group (P < 0.05).