Objective The diagnosis of delirium depends on eliciting its features through mental status examination and informant history. However, there is marked heterogeneity in how these features are assessed, from binary subjective clinical judgement to more comprehensive methods supported by cognitive testing. The aim of this article is to review the neuropsychological research in delirium and suggest future directions in research and clinical practice. Methods We reviewed the neuropsychological literature on formal assessment and quantification of the different domains in delirium, focusing on the core feature of inattention. Results Few studies have characterised and quantified the features of delirium using objective methods commonly employed in neuropsychological research. The existing evidence confirms that patients with delirium usually show impairments on objective tests of attention compared with cognitively intact controls and, in most cases, compared with patients with dementia. Further, abnormal level of arousal appears to be a specific indicator of delirium. The neuropsychological evidence base for impairments in other cognitive domains in delirium, including visual perception, language and thought processes, is small. Conclusions Delirium diagnosis requires accurate testing for its features, but there is little neuropsychological research examining the nature of these features, or evaluating the reliability, validity and discriminatory power of existing assessment processes. More research using the neuropsychological approach has enormous potential to improve and standardise delirium assessment methods of the individual features of delirium, such as inattention, and in developing more robust reference standards to enable greater comparability between studies. Copyright © 2017 John Wiley & Sons, Ltd.