Introduction: Post-operative delirium remains a significant problem, particularly in the older surgical patient. Previous evidence suggests that the provision of supplementary visual feedback about ones environment via the use of a mirror may positively impact on mental status and attention (core delirium diagnostic domains). We aimed to explore whether use of an evidence-based mirrors intervention could be effective in reducing delirium and improving post-operative outcomes such as factual memory encoding of the Intensive Care Unit (ICU) environment in older cardiac surgical patients. Methods: This was a pilot time-cluster randomized controlled trial at a 32-bed ICU, enrolling 223 patients aged 70 years and over, admitted to ICU after elective or urgent cardiac surgery from October 29, 2012 to June 23, 2013. The Mirrors Group received a structured mirrors intervention at set times (e.g., following change in mental status). The Usual Care Group received the standard care without mirrors. Primary outcome was ICU delirium incidence; secondary outcomes were ICU delirium days, ICU days with altered mental status or inattention, total length of ICU stay, physical mobilization (balance confidence) at ICU discharge, recall of factual and delusional ICU memories at 12 weeks, Health-Related Quality of Life at 12 weeks, and acceptability of the intervention. Results: The intervention was not associated with a significant reduction in ICU delirium incidence [Mirrors: 20/115 (17%); Usual Care: 17/108 (16%)] or duration [Mirrors: 1 (1-3); Usual Care: 2 (1-8)]. Use of the intervention on ICU was predictive of significantly higher recall of factual (but not delusional) items at 12 weeks after surgery (p = 0.003) and acceptability was high, with clinicians using mirrors at 86% of all recorded hourly observations. The intervention did not significantly impact on other secondary outcomes. Conclusion: Use of a structured mirrors intervention on the post-operative ICU does not reduce delirium, but may result in improved factual memory encoding in older cardiac surgical patients. This effect may occur via mechanisms unrelated to delirium, altered mental status, or inattention. The intervention may provide a new means of improving outcomes in patients at risk of post-ICU anxiety and/or Post-Traumatic Stress Disorder. Trial Registration: Clinicaltrials.gov identifier NCT01599689.