Interruption of the breathing gas to a ventilated anaesthetized patient due to accidental disconnection or anaesthesia system malfunction may have serious consequences if not detected quickly. A series of tests which covers the range of foreseeable mechanical problems was developed and used to test the performance of three breathing gas interruption monitors, two commercially available and one developed at Vancouver General Hospital. The tests were designed to evaluate the performance of monitors as installed on anaesthesia systems under a variety of failure conditions, including endotracheal tube disconnection with and without occlusion of the opening, kinks in the inspiratory and fresh gas hoses, disconnection of the fresh gas hose, leaks in the breathing circuit, excessive high or low pressure in the scavenging circuit, continuing high breathing circuit pressure, and kinks in the circuit pressure sensing hose. Ability to detect both significant changes in ventilation variables and faults existing at initiation of ventilation were also tested over a representative range of ventilator and patient variables using circle, coaxial and paediatric circuits. Only complete endotracheal tube disconnections with no obstruction of the opening were reliably detected by all three monitors. A commercial monitor with a single fixed-threshold alarm level also detected fresh gas interruptions in circle and adult coaxial circuits, but failed to alarm in response to any other fault condition. A monitor with selectable pressure thresholds and high, low, and continuing pressure limits detected just under half of the fault conditions. A microprocessor-based monitor developed at Vancouver General Hospital detected and correctly identified roughly 80 per cent of the faults. The series of tests forms the basis for a Canadian Standards Association Preliminary Standard (Z168.10) and will allow hospitals to test the performance of breathing gas interruption monitors in use in their institutions. Comments on the test series are solicited.