We have compared four tests for assessing airway sealing pressure with the laryngeal mask airway (LMA) to test the hypothesis that airway sealing pressure and inter-observer reliability differ between tests. We studied 80 paralysed, anaesthetized adult patients. Four different airway sealing pressure tests were performed in random order on each patient by two observers blinded to each other's measurements: test 1 involved detection of an audible noise; test 2 was detection of end-tidal carbon dioxide in the oral cavity; test 3 was observation of the aneroid manometer dial as the pressure increased to note the airway pressure at which the dial reached stability; and test 4 was detection of an audible noise by neck auscultation. Mean airway sealing pressure ranged from 19.5 to 21.3 cm H2O and intra-class correlation coefficient was 0.95-0.99. Inter-observer reliability of all tests was classed as excellent. The manometric stability test had a higher mean airway sealing pressure (P < 0.0001) and better inter-observer reliability (P < 0.0001) compared with the three other tests. We conclude that for clinical purposes all four tests are excellent, but that the manometric stability test may be more appropriate for researchers comparing airway sealing pressures.