We determined the effects of pretreatment with midazolam or fentanyl on the ease of laryngeal mask airway insertion using propofol without a muscle relaxant. One hundred and eighty ASA class I or II patients were randomly allocated to one of three groups, 60 patients per group, to receive either placebo (Group C), midazolam 0.05 mg kg-1 (Group M) or fentanyl 2 micrograms kg-1 (Group F), respectively. Following intravenous administration of these drugs, Group C received propofol 2.5 (Group C-2.5) or 3.0 mg kg-1 (Group C-3.0). Group M and Group F received propofol 2.0 or 2.5 mg kg-1 (Group M-2.0, M-2.5, F-2.0 and F-2.5, respectively). There was a smaller incidence of severe head and limb movements on LMA insertion in Group M-2.5, Group F-2.0 and Group F-2.5 than in Group C-2.5. Airway obstruction and inadequate jaw relaxation, which were occasionally recognized in Group C and Group F patients, were not observed in Group M-2.5. Overall the ease of LMA insertion was significantly better in Group M-2.0, Group M-2.5 and Group F-2.5 than in Group C-2.5, however, the blood pressure in Group F after LMA insertion was significantly lower than in Group M. We conclude that pretreatment with midazolam 0.05 mg combined with propofol 2.5 mg kg-1 provides safe and satisfactory conditions for LMA insertion.