Delirium is a common clinical phenomenon, often described as a disorder of consciousness. Delirium is commonly under recognised. The usual response to under recognition is to exhort practitioners to do a better job, but perhaps under recognition should instead be seen as a daily pragmatic challenge to how delirium is conceptualised. Here we retain the view that delirium is a disorder of consciousness, but propose a more multidimensional approach to this key feature. We argue that delirium can be recognised through evaluating arousal, attention and temporal orientation. We suggest that this approach can be validated by testing whether it leads to better delirium identification, accounts for the characteristic clinical disturbances, explains why delirium is common in the extreme age groups and why in later life its boundaries often blend with dementia.