We have compared size 3 and size 4 laryngeal masks in 30 females and size 4 and size 5 in 30 males for success rate of insertion, incidence of airleak and pressure exerted on the pharynx. First, the ex vivo volume-pressure relationship of the mask was obtained. Second, after insertion of a mask, the cuff was inflated with the recommended maximum volume of air and intracuff pressure measured. Third, the incidence of airleak during a steady airway pressure of 18 cm H2O was noted. Fourth, the cuff was deflated until it just prevented airleak, and cuff pressure was measured. The mask was removed, the other size was inserted and the same procedure repeated. At the end of operation, final in vivo and ex vivo pressures were measured. The pressure exerted on the pharynx was calculated as the difference between ex vivo and in vivo intracuff pressures. It was always possible to insert both sizes in both sexes. In females, airleak occurred in 15 patients with the size 3 and in five patients with the size 4 (difference: P = 0.005) and in males, 21 patients with the size 4 and in four patients with the size 5 (P < 0.001). Removal of air to the minimum effective volume significantly decreased intracuff pressure and pressure on the pharynx; on removal of the mask, pressures were similar to, or lower than, initial pressures. Therefore, a larger mask (size 4 in females and size 5 in males) provided a better seal than a smaller size without producing higher pressures on the pharynx.