Abstract Background Cognitive impairment often goes undetected in older people in hospital. Efficient screening tools are required to improve detection. To determine diagnostic properties of two separate informant-based single screening questions for cognitive impairment (dementia and delirium) in hospitalised older people. Methods Patients over 65 years non-electively admitted to medical or geriatric wards within a teaching hospital. Our index tests were single screening questions (SSQ), one for dementia (“How has your relative/friend’s memory changed over the past 5 years (up to just before their current illness)?”) and one for delirium (“How has your relative/friend’s memory changed with his/her current illness?”), which were assessed with informant response given on a five point Likert scale. Any deterioration on our index tests of SSQ-dementia and SSQ-delirium was accepted as a positive screen for cognitive impairment. Scores were compared to the Informant Questionnaire for Cognitive Decline in the Elderly (IQCODE) > 3.38 accepted as dementia, and Confusion Assessment Method (CAM) diagnosis of delirium. We also collected direct cognitive screening data using Mini Mental Status Examination (MMSE). Results Informant responses were obtained in 70/161 (43.5%) patients, median age 80.8 (range:67–97) years; mean MMSE score 18.5 (SD: 8.1). The SSQ-dementia when compared to the IQCODE had a sensitivity of 83.3% and specificity of 93.1%. The SSQ-delirium when compared to CAM diagnosis had sensitivity of 76.9% and a specificity of 56.1%. Conclusions These findings show promise for use of an informant single screening question tool as the first step in detection of dementia in older people in acute hospital care, although this approach appears to be less accurate in screening for delirium.