OBJECTIVE To assess the ease of insertion of laryngeal mask airway (LMA) comparing propofol with lidocaine or midazolam followed by thiopentone and compare the costs with each technique. METHODS One hundred and fifty ASA 1 or 2 patients equally divided into three groups scheduled for elective surgery were recruited into this prospective, single blind, randomized, parallel groups study. Anaesthetic induction was achieved with 1 microgram.kg-1 fentanyl i.v. followed by either 2.5 mg.kg-1 propofol (group P), or a sequence of 1.5 mg.kg-1 lidocaine and 5 mg.kg-1 thiopentone (group LT), or midazolam 0.1 mg.kg-1 and, three minutes later, 5 mg.kg-1 thiopentone (group MT). The LMA was inserted by the blinded anaesthetist who assessed and graded the conditions for LMA insertion and noted any adverse responses (i.e., inadequate jaw relaxation, gagging, coughing, limb or head movement, hiccough and laryngospasm). Conditions were considered "excellent" if there were no adverse responses, and "satisfactory" if such a response was mild and transient. RESULTS Excellent or satisfactory conditions were observed in 48 (96%) patients in the midazolam-thiopentone group, 46 (92%) in the propofol group, and 34 (68%) in the lidocaine-thiopentone group (P = 0.0001). The incidence of gagging (P = 0.042), limb movement (P = 0.031), and laryngospasm (P = 0.0001) was higher in the lidocaine-thiopentone group. CONCLUSIONS With the above doses, a fentanyl-midazolam-thiopentone combination which is about 35% less expensive than fentanyl-propofol, provides equally good conditions for the insertion of LMA.