Objectives: Delirium is common in older patients admitted to hospital. Information obtained from patient's relatives or caregivers may contribute to improved detection. Our aim was to develop a caregiver based questionnaire, the Informant Assessment of Geriatric Delirium (I-AGeD), to assist in better recognition of delirium in elderly patients. Methods: A cross-sectional observational study using a scale construction patient cohort and two validation cohorts was conducted at geriatric departments of two teaching hospitals in the Netherlands. Delirium status, based on DSM-IV criteria, was assessed directly on admission by a geriatric resident and research coordinator and evaluated within the first 48 hours of admission. Questionnaire item sampling was based on discussions with an expertpanel. Caregivers filled out a 37-item questionnaire of which ten items were selected reflecting delirium symptoms, based on their discriminatory abilities, internal consistency and interitem correlations. Results: A total of 88 patients with complete study protocols in the construction cohort were included. Average age was 86.4 (SD 8.5), and 31/88 patients had delirium on admission. Internal consistency of the 10-item I-AGeD was high (Cronbach's alpha = 0.85). At a cut-off score greater than 4 sensitivity was 77.4% and specificity 63.2%. In patients without dementia, sensitivity was 100% and specificity 65.2%. Validation occurred by means of two validation cohorts, one consisted of 59 patients and the other of 33 patients. Sensitivity and specificity in these samples ranged from 70.0-88.9% and 66.7%-100%. Conclusion: The newly constructed caregiver based I-AGeD questionnaire is a valid screening instrument for delirium on admission to hospital in geriatric patients.