The first 2000 incidents reported to the Australian Incident Monitoring Study were analysed with respect to the role of the capnograph. One hundred and fifty-seven (8%) were first detected by a capnograph and there were a further 18 (1%) in which capnography was contributory. Of the 1256 incidents which occurred in association with general anaesthesia 48% were "human detected" and 52% "monitor detected". The capnograph was ranked second and detected 24% of these monitor detected incidents; this figure would have been nearly 30% if a correctly checked, calibrated capnograph had always been used. The capnograph is a "front-line" monitor for oesophageal intubation, failure of ventilation, anaesthetic circuit faults, gas embolism, sudden circulatory collapse and malignant hyperthermia. It is a valuable "back-up" monitor when other monitors (e.g. low pressure alarm, pulse oximeter) are not in use, are being used incorrectly or fail. Such situations, in order of frequency of detection were: circuit-leak, overpressure of the breathing circuit, bronchospasm, leak of ventilator-driving-gas into the patient circuit, aspiration and/or regurgitation and hypoventilation. There were 20 reports of "failure", over two-thirds of which would not have occurred with appropriate checking and calibration. Seven were due to gas sampling problems and 6 to apnoea alarm failure. Two circuit leaks and 2 faulty unidirectional valves were not detected; on 3 occasions problems occurred due to power failure, calibration problems, or misinterpretation of an alarm.(ABSTRACT TRUNCATED AT 250 WORDS)