Objectives: To determine whether a bundled risk screening and warning or action card system improves formal delirium diagnosis and person-centered outcomes in hospitalized older adults. Design: Prospective trial with sequential introduction of screening and interventional processes. Setting: Two tertiary referral hospitals in Australia. Participants: Individuals aged 65 and older presenting to the emergency department (ED) and not requiring immediate resuscitation (N = 3,905). Intervention: Formal ED delirium screening algorithm and use of a risk warning card with a recommended series of actions for the prevention and management of delirium during the subsequent admission. Measurements: Delirium diagnosis at hospital discharge, proportion discharged to new assisted living arrangements, in-hospital complications (use of sedation, falls, aspiration pneumonia, death), hospital length of stay. Results: Participants with a positive risk screen were significantly more likely (relative risk = 6.0, 95% confidence interval = 4.9–7.3) to develop delirium, and the proportion of at-risk participants with a positive screen was constant across three study phases. Delirium detection rate in participants undergoing the final intervention (Phase 3) was 12.1% (a 2% absolute and 17% relative increase from the baseline rate) but this was not statistically significant (P =.29), and a similar relative increase was seen over time in participants not receiving the intervention. Conclusion: A risk screening and warning or action card intervention in the ED did not significantly improve rates of delirium detection or other important outcomes.