Background Delirium is an acute neuropsychiatric syndrome characterised by the rapid onset of severe and fluctuating disturbances in attention, arousal and cognition. Delirium is associated with a range of adverse outcomes; however, it goes undetected in up to 75% of cases. A light-counting task implemented on a custom-built computerised device entitled the Edinburgh Delirium Test Box (EDTB) has recently been validated as an objective screening tool for detecting inattention in delirium. Although the EDTB task is effective, tests contained on readily available devices have the potential for far greater impact. The current study evaluated the feasibility and validity of a smartphone adaptation of the EDTB known as the DelApp. Methods Hospital in-patients aged 60–96 were recruited for three groups: (1) patients with no delirium or known cognitive impairment (n=54); (2) patients with dementia but no current delirium (n=52); and (3) patients with delirium (n=50). Total DelApp score comprises an arousal scale and a light-counting task capturing sustained attention deficits in delirium. Additional tests included: conventional tests of inattention, the Short Orientation-Memory-Concentration test, the Delirium Rating Scale–Revised-98, the Confusion Assessment Method and the Observational Scale of Level of Arousal. Results Delirium patients (mdn=6, IQR=4-7) performed significantly worse than those with dementia (mdn=10, IQR=9-10; p<.001, r=.76) and controls (mdn=10, IQR=10-10; p<.001, r=.88). DelApp scores also correlated with symptom severity (τ =-.57, p<.001). General cognitive impairment did not differ between the delirium and dementia groups and therefore could not account for sustained attention deficits. Receiver operating characteristic analysis indicated that the DelApp was 98% sensitive and 93% specific to delirium. Discussion These results, like those of the EDTB, highlight the utility of the objective assessment of attention to identify delirium and differentiate it from dementia. The DelApp could be cheaply and easily distributed among healthcare professionals and has promising utility in clinical and research settings.