Delirium is a common and serious complication of acute illness in people admitted to hospital. Recent evidence suggests that delirium is associated with an increased subsequent risk of developing dementia, although it is unclear whether delirium merely serves as a marker of a preexisting risk or whether it may of itself provoke or facilitate development of permanent brain damage (1). Despite the importance of delirium, relatively little is known about the neuropathogenesis of the disorder. The main hypothesis at present is that delirium represents the clinical manifestation of diffuse, reversible impairment of neurotransmission and of cerebral oxidative metabolism. This hypothesis explains the relatively stereotyped nature of the response to a wide range of insults and the characteristic generalized slowing of the EEG trace in most cases of delirium. However, there is little direct evidence of impaired cerebral oxidative metabolism in delirium. The finding of delirium as a feature of some strokes suggests that more focal disturbances can also cause delirium. Conflicting findings have been reported from the few studies using positron emission tomography or single photon emission tomography in delirious patients. Various neurotransmitter systems have been implicated in the pathogenesis of delirium. The cholinergic hypothesis has attracted most attention. This proposes that reduced synthesis of acetylcholine is a final common pathway for development of delirium. Anticholinergic drugs are a potent cause of delirium, and anticholinergic toxicity is an established animal model for delirium at present; also, anticholinergic pathways are involved in regulation of attentiveness, disruption of which is a cardinal feature of delirium. However, there is a wide number of other candidate neurotransmitters including dopamine, serotonin GABA, glutamate, opioids and noradrenaline, and it is unlikely that changes in a single neurotransmitter can explain the protean features and aetiological factors of delirium. The existence of distinct behavioural subtypes may provide additional clues to the pathogenesis of delirium. Two characteristic subtypes are described: an agitated (hyperactive-hyperalert) variant, typified by delirium tremens, and a retarded (hypoactive-hypoalert) variant, typified by the metabolic encephalopathies. However, many patients switch between hypoactivity and hyperactivity and many aetiological factors, such as infection, can cause either variant. Also, although delirium tremens is classically associated with EEG fast wave activity, slow wave activity is more characteristic of agitated delirium due to anticholinergic medications or traumatic brain injury. Nevertheless, different subtypes may involve different neurotransmitter pathways, and require different treatments. It has been proposed that delirium might be a stress reaction in the elderly due to increased levels of glucocorticoids or an increased vulnerability to their effects (2). Hypothalamic-pituitary-adrenal (HPA) activation is related to cognitive impairment in depression and dementia, and increased glucocorticoid levels can cause hippocampal damage. Cushing’s syndrome and exogenous glucocorticoids can cause delirium (as can sudden cessation of glucocorticoid therapy). Glucocorticoids cause prominent deficits in selective attention; such deficits are a characteristic feature of delirium but not usually of dementia. In demented patients, the greatest slow wave EEG activity occurs in patients with highest resting cortisol levels. A number of studies have now demonstrated HPA activation in delirious patients. For example, in stroke patients, post-dexamethasone cortisol levels were associated with delirium, independently of the site and severity of the stroke (3). It remains unclear whether hypercortisolaemia is of pathogenetic significance in delirious patients or is simply a result of the stress of delirium or of the underlying illness. Even if HPA activation results from rather than causes delirium, it is possible that excessive cortisol secretion might mediate damage to hippocampal neurons thereby producing permanent brain damage and cognitive loss. Other interactions between the nervous and immune systems have also been implicated in the pathogenesis of delirium. Interleukin-1 released from macrophages or microglia can induce activation of