Establishing the diagnosis of brain death has never been easy for most physicians. Recently it has become even more complex as modern technology allows doctors to prolong almost indefinitely cardiac and respiratory function in comatose patients. In addition, because of advances in transplant medicine, it is more crucial than ever to make a diagnosis of brain death quickly and accurately in order to harvest transplant organs before they are damaged. The field of nuclear medicine could play an important part in this rapidly evolving area by providing a proven, sensitive and specific test to help make the diagnosis. Most Western European countries except Denmark now agree that 'brain death' equals 'death' (Danish Ethics Council 1990). Gone are the days when 'death' meant 'cessation of all organ systems'. Some writers prefer the more cumbersome term 'irreversible cessation of brain function' (ICBF) to remind doctors exactly what they are deciding when they form the diagnosis (George et al. 1990). At present, the diagnosis of brain death is largely clinical. Numerous professional committees have met and proposed guidelines in this area (Beecher 1968; Mohandas and Chou 1971; President's Commission 1981; Task Force 1987). The key elements in an adult are: (1) the absence of cerebral functions; (2) the absence of brain stem functions including spontaneous respirations for a period of at least 6 h; and (3) an irreversible state (Adams and Victor 1985). Cerebral function is evaluated by looking for volitional activity and checking the patient's reaction to auditory (shout), visual (wave) and cutaneous (pinch) stimuli (Adams and Victor 1985). Some movements, particularly spinal reflexes, may occur in a brain-dead patient (Ropper 1984; Jordan et al. 1985). There is even a 'Lazarus sign', where patients, 5-10 min after the ventilator has been stopped, show sudden movements of the arms and shoulders (Jordan et al. 1985). One patient, 5 min after respiratory support was stopped and cardiac function ceased, crossed both arms over his chest and began to sit up (Ropper 1984). Mechanical and hypoxic stimuli probably trigger upper cervical cord motor neurons to produce this sign. Brain stem functions are commonly evaluated by testing pupillary reaction to light, reaction to corneal stimulation, and the oculocephalic, oculovestibular, oropharyngeal and gag reflexes (Plum and Posner 1982). The most essential test is the apnea test, to see whether or not high levels of carbon dioxide can stimulate the respiratory portion of the brain stem. If the patient does not breathe after 10 min with documented levels of PCO2>60 mmHg, then the respiratory drive function of the brain stem is absent (Outwater and Rockoff 1984). In addition to searching for possible cerebral and brain stem function, one should try and establish an aetiology of the coma. If the cause is known and irreversible, one examination alone in an adult is sufficient to declare brain death. If the cause of the coma is unknown, then one must exclude possible reversible causes. Common examples of reversible coma include sedative-hypnotic agents (benzodiazepines, phenobarbital), paralytic agents, toxic and metabolic disorders (hepatic encephalopathy, uraemia, porphyria), hypothermia and hypotension (Plum and Posner 1982). Most authors recommend that for an adult, if the cause of a coma is unknown, reversible causes should be excluded and two examinations should be performed 12 h apart before making the diagnosis (Mohandas and Chou 1971; President's Commission 1981 ; Task Force 1987; George et al. 1990). The essential problem in establishing brain death, after showing that no cerebral or brain stem function exists, lies in determining irreversibility. In an adult with a known cause of coma this is not a problem. However, in newborns, infants and adults with coma of undetermined aetiology, it is often difficult to prove irreversibi-